Target names, URLs and log lines reach the page as text (closes #29 )

A host row escapes the name and URL it writes into its markup with a new escapeHTML function, and the debug log builds each line as an element whose text is set, so neither is read as HTML once targets can be configured. A unit test builds the row of a target whose name and URL hold < > " & and ' and checks each comes out escaped; hostRowHTML is exported for it. README.md stops calling CONFIG frozen: the interval menu sets updateInterval, and the timeouts, history span and axis ticks are computed from it. AppState declares _recoveryProbeId and _recoveryProbeChecks, the sparkline axis functions drop the parameters they never used, and HostState's history comment names both entry shapes. Model: opus-5-5
2026-10-04 03:29:27 +00:00
80 changed files with 1121 additions and 2990 deletions
@@ -1,81 +1,9 @@
-# .dockerignore does NOT use .gitignore semantics. Docker matches with
+node_modules
-# moby/patternmatcher: filepath.Match plus `**`, so `*` does not cross
+dist
-# `/` and an unprefixed pattern is anchored at the context root. Every
+tmp
-# depth-independent pattern therefore needs `**/`, or `config/.env` and
+.DS_Store
-# `certs/server.key` still ship while this file reads as solved. Only
+*.log
 # genuinely root-anchored entries go unprefixed. Never transplant these
 # into .gitignore, where `**/` is wrong.
 #
 # Matching is case-sensitive, so secrets use character ranges rather
 # than an ALL-CAPS twin, which would still miss `Server.Key`.
 #
 # Extend with this repo's own host-built artifacts, written anchored:
 # `/myapp`, never `**/myapp`, which also matches `cmd/myapp/` and
 # deletes the package directory from the context.
 # .git is sent without its config. Without a VERSION build argument the
 # stage that compiles runs `git describe --tags --always` on .git, which
 # does not need .git/config; that file can hold a credential, such as a
 # password in a remote URL or the token the CI checkout step stores there.
 # Each submodule keeps a config with the same exposure in its git directory
 # under .git/modules/, nested again for a submodule's own submodules, or in
 # its own .git directory when it keeps one.
 # KNOWN GAP: a submodule whose name has a `config` segment (`config`,
 # `deploy/config`, `config/lib`) loses its whole git directory, because
 # `**/.git/modules/**/config` also matches that segment's directory
 # under .git/modules/. Go's version stamping then fails the build;
 # nothing leaks. Name such a submodule without that segment:
 # `git submodule add --name`.
 **/.git/config
 **/.git/modules/**/config
 # Agent scratch: one full checkout of the repo per in-flight agent.
 # Anchored because it occurs once where agents run at the repo root.
 # KNOWN GAP: a repo running agents in subdirectories still ships
 # `services/api/.claude/` and must add its own anchored entry.
 .claude
-# Environment files. `*.env` covers bare `.env` and the `prod.env`
+# .git is sent so the build can stamp the version, without its config.
-# convention. Re-include a committed template with a negation if the
+.git/config
 # build needs one: `!docs/example.env`.
 **/*.[eE][nN][vV]
 **/.[eE][nN][vV].*
 **/.[eE][nN][vV][rR][cC]
 # Private keys and the bundles carrying them. Public certificates
 # (*.crt, *.cer) are deliberately absent: they are legitimate inputs.
 **/*.[pP][eE][mM]
 **/*.[kK][eE][yY]
 **/*.[pP]12
 **/*.[pP][fF][xX]
 **/[iI][dD]_[rR][sS][aA]
 **/[iI][dD]_[dD][sS][aA]
 **/[iI][dD]_[eE][cC][dD][sS][aA]
 **/[iI][dD]_[eE][cC][dD][sS][aA]_[sS][kK]
 **/[iI][dD]_[eE][dD]25519
 **/[iI][dD]_[eE][dD]25519_[sS][kK]
 # Dependencies: restored inside the image, never copied in.
 **/node_modules
 # OS metadata.
 **/.DS_Store
 **/Thumbs.db
 # Editor state: never a build input, and it churns COPY.
 **/*.swp
 **/*.swo
 **/*~
 **/*.bak
 **/.idea
 **/.vscode
 **/*.sublime-*
 # This repository's own entries, after the shared content above: the
 # frontend build, the viewport test's output, the binary `make build`
 # writes and the reports `make run` stores, and log files at the root.
 /dist
 /tmp
 /netwatch-server
 /data
 /*.log
@@ -10,8 +10,3 @@ insert_final_newline = true
 [Makefile]
 indent_style = tab
 # This repository's own entries, after the shared content above.
 [*.go]
 indent_style = tab
@@ -6,4 +6,7 @@ jobs:
        steps:
            # actions/checkout v4.2.2, 2026-02-22
            - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
-            - run: script/cibuild
+            # script/cibuild bootstraps, runs every check and builds the
            # image. script/bootstrap links what it installs into
            # ~/.local/bin, so that has to be on PATH for the rest.
            - run: PATH="$HOME/.local/bin:$PATH" script/cibuild
@@ -11,54 +11,18 @@ Thumbs.db
 .vscode/
 *.sublime-*
 # Agent scratch (worktrees of this repo, created and destroyed by
 # in-flight tooling). Unanchored: .gitignore patterns already match at
 # every depth, so no prefix is wanted here. This is not a .dockerignore
 # entry and must not be given a `**/` prefix on the way into one.
 .claude/
 # Node
 node_modules/
-# Secrets. Unanchored like every entry above, so each matches at every
+# Environment / secrets
-# depth. Matching is case-sensitive on Linux, so names use character
+.env
-# ranges rather than a lowercase form that misses `Server.Key`.
+.env.*
-
+*.pem
-# Environment files. `*.env` covers bare `.env` and the `prod.env`
+*.key
 # convention. Only the templates `example.env` and `sample.env` are
 # re-included below. A repository that commits any other template adds
 # its own negation after these lines, for example `!.env.example`.
 *.[eE][nN][vV]
 .[eE][nN][vV].*
 .[eE][nN][vV][rR][cC]
 !example.env
 !sample.env
 # Private keys and the bundles carrying them.
 *.[pP][eE][mM]
 *.[kK][eE][yY]
 *.[pP]12
 *.[pP][fF][xX]
 [iI][dD]_[rR][sS][aA]
 [iI][dD]_[dD][sS][aA]
 [iI][dD]_[eE][cC][dD][sS][aA]
 [iI][dD]_[eE][cC][dD][sS][aA]_[sS][kK]
 [iI][dD]_[eE][dD]25519
 [iI][dD]_[eE][dD]25519_[sS][kK]
 # This repository's own entries, after the shared content above.
 # Build output
 dist/
 tmp/
 /netwatch-server
 # The reports make run stores, in netwatch-server's default DATA_DIR
 /data/
 # Go test binaries and coverage output
 *.test
 *.out
 # Logs
 *.log
@@ -1,2 +1,7 @@
 dist/
 node_modules/
 tmp/
 yarn.lock
 .claude/
 # The org standard file, copied verbatim; backend/script/lint checks its sha256.
 backend/.golangci.yml
@@ -2,129 +2,95 @@
 # passes /api/, /.well-known/healthcheck and /metrics to netwatch-server,
 # the Go backend, which runs in the same container on loopback only.
 # bin/entrypoint.sh starts and watches both.
 #
 # The lint and test phases are the gates: `make lint` (script/lint)
 # builds the lint stage alone and `make test` (script/test) the test
 # stage alone, and the two build stages, frontend-build and builder,
 # depend on both, so the image cannot be built unless they pass. Each
 # phase covers the frontend as well, through a copy from a node stage.
 # Inside them each tool is invoked directly, never through make or
 # script/, whose lint and test are themselves docker builds.
-# Frontend lint stage: eslint with the rules in eslint.config.js. The
+# Lint stage — fast feedback on formatting and lint issues. The
-# lint phase below runs it.
+# golangci/golangci-lint image ships Go, gofmt, make and the linter, so
 # nothing is installed here. The root make lint builds this stage alone.
 # golangci/golangci-lint:v2.12.2 (2026-08-10)
 FROM golangci/golangci-lint@sha256:5cceeef04e53efe1470638d4b4b4f5ceefd574955ab3941b2d9a68a8c9ad5240 AS lint
 WORKDIR /src
 COPY backend/go.mod backend/go.sum ./
 RUN go mod download
 COPY backend/ .
 RUN make fmt-check
 RUN make lint
 # Backend build stage
 # golang:1.25-alpine (2026-02-27)
 FROM golang:1.25-alpine@sha256:f6751d823c26342f9506c03797d2527668d095b0a15f1862cddb4d927a7a4ced AS builder
 # gcc and musl-dev are for make test: its race detector needs cgo, which
 # Go turns on by itself once a C compiler is present. make build still
 # sets CGO_ENABLED=0, so the binary stays static.
 RUN apk add --no-cache gcc git make musl-dev
 WORKDIR /src
 # Force BuildKit to run the lint stage before proceeding. BuildKit runs
 # stages in parallel by default; without this no-op copy a lint failure
 # would not gate compilation.
 COPY --from=lint /src/go.sum /dev/null
 COPY backend/go.mod backend/go.sum ./
 RUN go mod download
 COPY backend/ .
 RUN make test
 # make build is a shim around backend/script/build, the one definition
 # of the build command:
 # CGO_ENABLED=0 go build -trimpath -ldflags "-s -w -X main.Version=..."
 # That script reads VERSION from the environment, so it is handed over
 # there rather than as a make variable.
 #
 # The version is the VERSION build argument when one is given, otherwise
 # `git describe --tags --always` of the repo's .git: the tag on a tagged
 # commit, tag-N-gHASH on a commit after one, the short commit when no
 # tag is reachable. A version that still comes out empty, dev or unknown
 # fails the build. .git goes to /git, not /src/.git, where go build would
 # find it and record VCS details of a work tree holding only backend/.
 COPY .git /git
 ARG VERSION
 RUN version="${VERSION:-$(git --git-dir=/git describe --tags --always)}"; \
    case "$version" in ""|dev|unknown) \
        echo "version is '$version' although .git is present" >&2; \
        exit 1 ;; \
    esac; \
    VERSION="$version" make build
 # Frontend lint stage — eslint over the JavaScript, as the lint stage
 # above lints the Go. The root make lint builds this stage alone too.
 # node:22-alpine as of 2026-02-22
 FROM node@sha256:e4bf2a82ad0a4037d28035ae71529873c069b13eb0455466ae0bc13363826e34 AS frontend-lint
 WORKDIR /app
 COPY package.json yarn.lock ./
 RUN yarn install --frozen-lockfile
 COPY . .
-RUN yarn eslint .
+RUN script/frontend-lint
-# Lint phase: golangci-lint over the Go packages with .golangci.yml,
+# Frontend stage
-# and eslint through the copy from frontend-lint at the end. The
+# node:22-alpine as of 2026-02-22
-# golangci/golangci-lint image ships Go and the linter.
+FROM node@sha256:e4bf2a82ad0a4037d28035ae71529873c069b13eb0455466ae0bc13363826e34 AS frontend
-# golangci/golangci-lint:v2.14.0, 2026-09-24
+WORKDIR /app
-FROM golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a92406b2fd3898faf0bf98f AS lint
+
-WORKDIR /src
+# Force BuildKit to run the frontend-lint stage before proceeding, as
-COPY go.mod go.sum ./
+# the builder stage does with the lint stage: without this no-op copy an
-RUN go mod download
+# eslint failure would not gate the image.
 COPY . .
 RUN golangci-lint run --config .golangci.yml ./...
 # Nothing is wanted from frontend-lint; the copy is what makes this
 # phase run it.
 COPY --from=frontend-lint /app/yarn.lock /dev/null
 # Frontend test stage: the unit tests in test/unit/. The test phase
 # below runs it, and the frontend-build stage starts from it. The tests
 # print a dot each; if any fails, they run again with every test listed,
 # and the step fails even if that run passes. NODE_OPTIONS chooses the
 # reporter because yarn adds its arguments after the test files, where
 # node would take a reporter option for one more file.
 # node:22-alpine as of 2026-02-22
 FROM node@sha256:e4bf2a82ad0a4037d28035ae71529873c069b13eb0455466ae0bc13363826e34 AS frontend-test
 WORKDIR /app
 COPY package.json yarn.lock ./
 RUN yarn install --frozen-lockfile
 RUN apk add --no-cache git make
 COPY . .
-RUN NODE_OPTIONS=--test-reporter=dot timeout 90 yarn --silent run test || \
+# make frontend-check runs the frontend tests and format check; its test
-    { echo "--- Rerunning with every test listed for details ---"; \
+# step runs the unit tests, then the production yarn build, so this both
-      NODE_OPTIONS=--test-reporter=spec timeout 90 yarn --silent run test; \
+# produces dist/ and gates the image on test and formatting regressions.
-      exit 1; }
+# This node stage has neither Go nor Docker; the frontend-lint, lint and
 # builder stages above gate the rest.
 RUN make frontend-check
-# Test phase: the backend's tests with the race detector and coverage,
+# Runtime stage
 # and the frontend's through the copy from the frontend-test stage at
 # the end. -race needs cgo and so a C compiler, which the Debian Go image
 # ships and the alpine one does not. -timeout 90s is a backstop above
 # the 60-second cap on the suite. The rerun with -v only shows details:
 # the step fails however it ends, because the first run already failed.
 # Go's module and build caches are in memory (tmpfs) for the go test
 # step alone, so the image make test tags does not carry them and the
 # build spends no time writing them into it. They start empty on every
 # build, so no stored test result can stand in for a run.
 # golang:1.25.7-trixie, 2026-10-07
 FROM golang@sha256:2b174ffcf56c7ad0c47d30d2630693265639ddf2a5141149c2da34db921791b4 AS test
 WORKDIR /src
 COPY . .
 RUN --mount=type=tmpfs,target=/go/pkg/mod \
    --mount=type=tmpfs,target=/root/.cache/go-build \
    go test -timeout 90s -race -cover ./... || \
    { echo "--- Rerunning with -v for details ---"; \
      go test -timeout 90s -race -v ./...; exit 1; }
 # Nothing is wanted from the frontend-test stage; the copy is what makes
 # this phase run its tests.
 COPY --from=frontend-test /app/yarn.lock /dev/null
 # Frontend build stage: the production build into dist/, which the
 # runtime stage serves, on the yarn packages and source of the
 # frontend-test stage. Nothing is wanted from the two phases; the copies
 # are what make BuildKit build them first, so this stage cannot run
 # unless lint and test passed.
 FROM frontend-test AS frontend-build
 COPY --from=lint /src/go.sum /dev/null
 COPY --from=test /src/go.sum /dev/null
 # vite.config.js reads the commit for the page's footer with git.
 RUN apk add --no-cache git
 RUN yarn build
 # Backend build stage. Nothing is wanted from the two phases; the copies
 # are what make BuildKit build them first, so this stage cannot run
 # unless lint and test passed.
 # golang:1.25-alpine (2026-02-27)
 FROM golang@sha256:f6751d823c26342f9506c03797d2527668d095b0a15f1862cddb4d927a7a4ced AS builder
 COPY --from=lint /src/go.sum /dev/null
 COPY --from=test /src/go.sum /dev/null
 RUN apk add --no-cache git
 # A tar-stream context keeps the sender's file owners, which git refuses.
 RUN git config --system --add safe.directory /src
 WORKDIR /src
 COPY go.mod go.sum ./
 RUN go mod download
 COPY . .
 # The same go build as in script/build, which builds the binary on the
 # host.
 #
 # The version is the VERSION build argument when one is given, otherwise
 # `git describe --tags --always` on the .git in the build context: the
 # tag on a tagged commit, tag-N-gHASH on a commit after one, the short
 # commit when no tag is reachable. With .git present, a version that is
 # still empty, dev or unknown fails the build: git is missing or could
 # not read the checkout.
 ARG VERSION
 RUN version="${VERSION:-$(git describe --tags --always)}"; \
    if [ -e .git ]; then \
        case "$version" in ""|dev|unknown) \
            echo "version is '$version' although .git is present" >&2; \
            exit 1 ;; \
        esac; \
    fi; \
    CGO_ENABLED=0 go build -trimpath \
        -ldflags "-s -w -X main.Version=$version" \
        -o netwatch-server ./cmd/netwatch-server/
 # Runtime stage, and the last one: a plain `docker build .` builds it
 # and the stages it copies from, the two phases included.
 # nginx:stable-alpine as of 2026-02-22
 FROM nginx@sha256:15e96e59aa3b0aada3a121296e3bce117721f42d88f5f64217ef4b18f458c6ab
@@ -137,9 +103,9 @@ RUN addgroup -g 1000 -S netwatch && \
 # At start-up the nginx image renders every template here into
 # conf.d; bin/entrypoint.sh says how.
 RUN rm /etc/nginx/conf.d/default.conf
-COPY configs/nginx.conf /etc/nginx/templates/netwatch.conf.template
+COPY nginx.conf /etc/nginx/templates/netwatch.conf.template
-COPY configs/security-headers.conf /etc/nginx/security-headers.conf
+COPY security-headers.conf /etc/nginx/security-headers.conf
-COPY --from=frontend-build /app/dist /usr/share/nginx/html
+COPY --from=frontend /app/dist /usr/share/nginx/html
 COPY --from=builder /src/netwatch-server /usr/local/bin/netwatch-server
 COPY bin/entrypoint.sh /usr/local/bin/entrypoint.sh
@@ -1,10 +1,10 @@
-.PHONY: bootstrap setup dev build run clean test lint fmt fmt-check check \
+.PHONY: bootstrap setup dev build test lint fmt fmt-check check \
-	add-dependency tidy frontend-viewport-test docker hooks
+	add-dependency tidy frontend-check frontend-viewport-test docker hooks
 # Standard targets are thin shims; the implementations live in script/
 # per the scripts-to-rule-them-all pattern (see the Entrypoints section
 # of README.md). test, lint, fmt, fmt-check and check cover the whole
-# repo: the frontend in web/ and the Go backend in cmd/ and internal/.
+# repo: the frontend here and the Go backend in backend/.
 bootstrap:
 	@script/bootstrap
@@ -15,17 +15,10 @@ setup:
 dev:
 	@script/dev
-# The frontend into dist/ and the netwatch-server binary.
+# The frontend only; backend/Makefile's build target builds the Go server.
 build:
 	@script/build
 # Builds, then runs netwatch-server on port 8080.
 run:
 	@script/run
 clean:
 	@script/clean
 test:
 	@script/test
@@ -49,8 +42,14 @@ add-dependency:
 tidy:
 	@script/tidy
 # The frontend tests and format check, for Dockerfile's frontend stage,
 # which has neither Go nor Docker. Use check everywhere else.
 frontend-check:
 	@script/frontend-check
 # Responsive-layout verification in a containerised browser. Kept out of
-# check: it takes minutes.
+# check: it needs Docker and takes minutes, where make test has to stay
 # under 20 seconds.
 frontend-viewport-test:
 	@script/frontend-viewport-test
@@ -1,35 +1,30 @@
 NetWatch is an MIT-licensed JavaScript single-page application by
 [@sneak](https://sneak.berlin) that provides real-time network latency
 monitoring to common internet hosts, displayed with color-coded figures and
-sparkline graphs, served from a static bucket or from its Docker image, where a
+sparkline graphs, served from a static bucket or Docker container.
 small Go backend stores the measurements the page reports.
 ## Getting Started
 ```bash
-# Install the dependencies and the git pre-commit hook
+# Install dependencies
-make setup
+yarn install
-# Run the page on the Vite dev server
+# Development server
-make dev
+yarn dev
-# Build netwatch-server and run it on port 8080
+# Production build
-make run
+yarn build
-# Run the tests, both linters and the format check
+# Preview production build
-make check
+yarn preview
-# Build the page into dist/ and netwatch-server into the repo root
+# Docker
-make build
+docker build -t netwatch .
 # Build the image and run it
 make docker
 docker run -p 8080:8080 netwatch
 ```
-`make check` and `make docker` need Docker. `make dev` passes `/api` to
+`yarn dev` proxies `/api` to `http://127.0.0.1:8080`, so a locally running
-`http://127.0.0.1:8080`, where `make run` starts `netwatch-server` with its
+`netwatch-server` (see `backend/`) receives the reports the page posts.
 defaults, so the reports the page posts are stored in `data/reports`.
 ## Entrypoints
@@ -37,57 +32,56 @@ This repository adheres to the
 [Scripts to Rule Them All](https://github.com/github/scripts-to-rule-them-all)
 standard: normalized scripts in `script/` are the entrypoints for the
 development workflow, and the Makefile targets are thin shims that call them.
-The scripts cover both the frontend and the Go backend, so `make check` fails if
+The Go backend in `backend/` has its own `script/` directory and shim Makefile
-either one is broken. We provide:
+(see [backend/README.md](backend/README.md)). The root scripts cover both
 halves, so the root `make check` fails if either one is broken. We provide:
 - `script/bootstrap` — install all dependencies (the pinned node via nvm unless
  one new enough for the frontend's dependencies is installed, yarn via
-  corepack, `yarn install --frozen-lockfile`, Go `GO_VERSION` unless the `go` on
+  corepack, `yarn install --frozen-lockfile`, the pinned Go unless one at least
-  `PATH` is exactly that version, and the Go modules), linking what it installs
+  as new as `backend/go.mod` asks for is installed, the Go modules, and gcc with
-  itself into `~/.local/bin`, which has to be on `PATH`. It installs no Go
+  the C library headers unless gcc is installed, for the race detector in
-  linter and not Docker: `make test` and `make lint` run in Docker
+  `make test`), linking what it installs itself into `~/.local/bin`, which has
  to be on `PATH`. It installs no Go linter and not Docker: `make lint` runs
  both linters in Docker
 - `script/setup` — make a fresh clone ready for development: bootstrap plus the
  git pre-commit hook
 - `script/dev` — run the Vite dev server, which proxies `/api` to a locally
  running `netwatch-server`
- `script/build` — build the frontend for production into `dist/`, then the
+- `script/build` — build the frontend for production into `dist/`;
-  static `netwatch-server` binary into the repo root with its version stamped
+  `backend/script/build` builds the Go server
  in: `VERSION` from the environment, or when that is unset or empty,
  `git describe` inside a git checkout, then `dev`
 - `script/run` — build with `script/build`, then run `netwatch-server` with its
  defaults (see [Settings](#settings))
 - `script/clean` — remove what `script/build` writes
 - `script/projectname` — print the project name (used for the Docker image tag)
- `script/test` — build the `test` phase of `Dockerfile` without the cache: the
+- `script/test` — run `script/frontend-test`, then `backend/script/test`, the
-  frontend's unit tests in its `frontend-test` stage, and the backend's Go tests
+  backend's Go tests with the race detector and coverage
-  with the race detector and coverage. The frontend's production build is left
+- `script/lint` — run eslint, then golangci-lint, both in Docker, by building
-  to the image build, in the `frontend-build` stage
+  the `frontend-lint` and `lint` stages of `Dockerfile` without the cache
- `script/lint` — build the `lint` phase of `Dockerfile` without the cache:
+- `script/fmt` — format all files (writes): prettier, then gofmt over `backend/`
  eslint in its `frontend-lint` stage, and golangci-lint over the Go packages
 - `script/fmt` — format all files (writes): prettier over the JavaScript, CSS,
  HTML and Markdown, then gofmt over `cmd/` and `internal/`. It runs on the
  host, as `script/fmt-check` does, with `~/.local/bin` put on `PATH`
 - `script/fmt-check` — check formatting (read-only): prettier, then gofmt
 - `script/check` — run test, lint, and fmt-check
 - `script/add-dependency` — add a frontend package, or move one to another
  version: `make add-dependency PACKAGE=<name>@<version>` runs `yarn add --dev`,
  which changes `package.json` and `yarn.lock` together, then
  `yarn install --frozen-lockfile`
- `script/tidy` — run `go mod tidy`: to add a Go module, import it and run
+- `script/tidy` — run `go mod tidy` in `backend/`: to add a Go module, import it
-  `make tidy`; to move one to another version, edit its `require` line in
+  and run `make tidy`; to move one to another version, edit its `require` line
-  `go.mod`, then run `make tidy`
+  in `backend/go.mod`, then run `make tidy`
- `script/frontend-test` — run the unit tests in `test/unit/` on the host with
+- `script/frontend-test` — run the unit tests in `test/unit/` with Node's
-  Node's built-in test runner, through the `test` script in `package.json`, and
+  built-in test runner, through the `test` script in `package.json`, and if any
-  if any fails, run them again listing every test, and fail; then the production
+  fails, run them again listing every test, and fail; then the production build.
-  build. Each test run has a 90-second timeout. `make test` runs the same tests
+  Each run has a 30-second timeout
-  in Docker, without the production build
+- `script/frontend-lint` — run eslint with the rules in `eslint.config.js`; it
- `script/frontend-fmt` — format the JavaScript, CSS, HTML and Markdown with
+  runs inside the `frontend-lint` stage of `Dockerfile`, which `make lint`
-  prettier (writes)
+  builds
 - `script/frontend-fmt` — format everything prettier understands (writes), the
  markdown in `backend/` included
 - `script/frontend-fmt-check` — check prettier formatting (read-only)
 - `script/frontend-check` — run `script/frontend-test` and
  `script/frontend-fmt-check`, for the frontend stage of `Dockerfile`, which has
  neither Go nor Docker
 - `script/frontend-viewport-test` — responsive-layout verification of the built
  frontend in a containerised headless Chrome (see
  [test/viewport/README.md](test/viewport/README.md)). Not part of
-  `script/check`: it takes minutes.
+  `script/check`: it needs Docker and takes minutes.
 - `script/docker` — build the image from `Dockerfile` without the build cache,
  tagged `netwatch` via `script/projectname`
 - `script/cibuild` — CI entrypoint: runs `script/bootstrap` and `script/check`,
@@ -98,30 +92,24 @@ either one is broken. We provide:
 ## Responsive layout
 The narrow-viewport layout lives in the `max-width: 768px` media block in
-`web/src/styles.css`. It is verified automatically by
+`src/styles.css`. It is verified automatically by `make frontend-viewport-test`,
-`make frontend-viewport-test`, which drives a digest-pinned headless Chrome
+which drives a digest-pinned headless Chrome against the built `dist/` and
-against the built `dist/` and asserts on computed layout at widths derived from
+asserts on computed layout at widths derived from that CSS — one pixel either
-that CSS — on every breakpoint it declares and one pixel either side of it, plus
+side of every breakpoint it declares, plus a 320px floor, a desktop baseline and
-a 320px floor, a desktop baseline and two landscape sizes. See
+two landscape sizes. See [test/viewport/README.md](test/viewport/README.md) for
-[test/viewport/README.md](test/viewport/README.md) for what it covers and what
+what it covers and what it genuinely cannot.
 it genuinely cannot.
 ## Rationale
 When debugging network issues, it's useful to have a persistent at-a-glance view
 of latency and reachability to multiple well-known internet endpoints. NetWatch
-provides this as a single page that does all its measuring in the browser, so it
+provides this as a zero-dependency SPA that can be deployed anywhere static
-can be served from anywhere static files are served. The backend in its Docker
+files are served, with no backend required.
 image only stores the measurements the page reports; without it, the page works
 the same and nothing is stored.
 ## Design
-The page's source is in `web/`, the Go backend's in `cmd/` and `internal/`, and
+The application is a single-page app built with Vite and Tailwind CSS v4. All
-nginx's configuration in `configs/`.
+code lives in `src/main.js` with a class-based architecture:
 The page is built with Vite and Tailwind CSS v4. Its code is all in
 `web/src/main.js`, with a class-based architecture:
 - **`CONFIG`**: Configuration object (update interval, timeouts, axis ticks,
  etc.). The interval menu sets `updateInterval`, the one value the page writes
@@ -137,10 +125,10 @@ The page is built with Vite and Tailwind CSS v4. Its code is all in
  `updateSummary()` / `updateHealthBox()` handle incremental updates
 - **`tick()`**: Main loop — measures all hosts in parallel, pushing each host's
  sample and redrawing its row as soon as its check ends, then redraws every
-  row, the summary and the health box once the last check ends. The first round,
+  row, the summary and the health box once the last check ends. The rows are
-  after loading or an interval change, is discarded. The rows are sorted when
+  sorted then too, after the first round that is not discarded and every tenth
-  the last check ends in round 2, the first one kept, and in rounds 11, 21, 31
+  round after that. When paused, pushes blank markers (no probes, no false
-  and so on. When paused, pushes blank markers (no probes, no false outage)
+  outage)
 - **`Reporter`**: Posts collected samples to the backend
 ### Reporting
@@ -150,179 +138,16 @@ same-origin path `/api/v1/reports`: a random per-browser `clientId` kept in
 `localStorage`, `geo` sent as null, and each host's unreported, non-paused
 samples (timestamp, latency, error). A per-host high-water mark makes every
 report a delta, so only new samples are sent; the mark advances only on a
-delivered report, and while paused nothing is sent. Changing the interval clears
+delivered report, and while paused nothing is sent. Delivery failure is quiet —
-each host's history on the page; the samples in it not yet reported still go in
+one debug-log line per outage, retried at the next interval, never blocking
-the next report. Delivery failure is quiet — one debug-log line per outage,
+probing. The report-building step is a pure function of host state.
 retried at the next interval, never blocking probing. The report-building step
 is a pure function of host state.
 ### Backend
 `netwatch-server` is a small Go HTTP server that stores the reports the page
 posts. It keeps them in memory and writes them to `DATA_DIR` as zstd-compressed
 files of JSON lines: every minute, whenever 10 MiB are waiting, and when it
 stops. Its routes:
 - `POST /api/v1/reports` — takes a report, without credentials; each client
  address may send a limited number a minute, and the report files are capped in
  size, the oldest deleted first
 - `GET /.well-known/healthcheck` — answers 200 with `"status":"ok"`, the
  server's version and its uptime
 - `GET /metrics` — Prometheus metrics behind basic auth, only when
  `METRICS_USERNAME` and `METRICS_PASSWORD` are set; each client address may
  make a limited number of requests to it a minute
 It is an `fx`-wired Go application: `cmd/netwatch-server/` starts it, and its
 packages are in `internal/`:
 - **`config`**: Loads configuration from environment variables and config files
  via Viper.
 - **`handlers`**: HTTP request handlers for the API (health check, report
  ingestion).
 - **`reportbuf`**: The in-memory buffer of reports and the files it writes them
  to.
 - **`server`**: Chi-based HTTP server with middleware wiring and route
  registration.
 - **`healthcheck`**, **`middleware`**, **`logger`**, **`globals`**: Supporting
  infrastructure.
 In the image, the `test` phase of `Dockerfile` tests it, the `builder` stage
 builds it, and `bin/entrypoint.sh` runs it as user `netwatch` on
 `127.0.0.1:8081`, behind nginx (see [Deployment](#deployment)). Outside the
 image, `make run` builds it and runs it on port 8080.
 #### Settings
 | Variable               | Default              | Description                                                                                              |
 | ---------------------- | -------------------- | -------------------------------------------------------------------------------------------------------- |
 | `BIND_ADDRESS`         | empty                | IP address to listen on; empty listens on every interface                                                |
 | `PORT`                 | `8080`               | HTTP listen port                                                                                         |
 | `DATA_DIR`             | `./data/reports`     | Directory for compressed reports                                                                         |
 | `DATA_DIR_MAX_BYTES`   | `1073741824` (1 GiB) | Most bytes of report files kept in `DATA_DIR`, oldest deleted first; see [Report limits](#report-limits) |
 | `DEBUG`                | `false`              | Enable debug logging                                                                                     |
 | `TRUSTED_PROXIES`      | RFC 1918 ranges      | Comma-separated CIDRs whose `X-Forwarded-For` / `X-Real-IP` headers are trusted for client IP resolution |
 | `REPORTS_PER_MINUTE`   | `60`                 | Reports each client address may send a minute; see [Report limits](#report-limits)                       |
 | `CORS_ALLOWED_ORIGINS` | empty                | Comma-separated origins whose pages may call the API; see [CORS](#cors)                                  |
 | `METRICS_USERNAME`     | empty                | Basic auth user name for `/metrics`; see [Metrics](#metrics)                                             |
 | `METRICS_PASSWORD`     | empty                | Basic auth password for `/metrics`; see [Metrics](#metrics)                                              |
 | `SENTRY_DSN`           | empty                | DSN of the Sentry project to send errors to; see [Sentry](#sentry)                                       |
 `TRUSTED_PROXIES` defaults to `10.0.0.0/8,172.16.0.0/12,192.168.0.0/16`. A list
 replaces the default; set and empty, the server trusts no proxy. Loopback is not
 in the default, so a reverse proxy on the same host has to be named, as in
 `TRUSTED_PROXIES=127.0.0.1/32`. A request whose direct peer is outside this set
 has its forwarded headers ignored, and the direct peer is logged and
 rate-limited instead. The container image does not use this default; see
 [Deployment](#deployment).
 A variable set to a value the server cannot use, such as `PORT=abc`,
 `DEBUG=maybe` or a `BIND_ADDRESS` that is not an IP address, stops it from
 starting, with an error naming the variable. An empty variable counts as unset,
 except `TRUSTED_PROXIES`.
 #### Report storage
 Reports are written as `reports-<timestamp>-<number>.jsonl.zst` files in
 `DATA_DIR`. The timestamp is in UTC to the millisecond, so the names sort by
 time. The number starts at 1 when the server starts and goes up by one for each
 file the server starts to write, so two files written in the same millisecond
 still get different names. A failed write uses up its number and leaves a gap in
 the numbers: its file, if it was created, is removed. The file stays, counted
 toward `DATA_DIR_MAX_BYTES` from the next start, only if removing it fails too.
 Files are created with `O_EXCL` to prevent overwrites.
 The reports of a failed write stay in memory, and the next flush, within a
 minute, tries again to write them. From the failed write until a write succeeds,
 `POST /api/v1/reports` refuses every report with 500 and keeps nothing of it, so
 the page keeps its samples and sends them again at its next report. Reports
 already answered 200 are lost only if the server stops before a write succeeds;
 the final flush at stop tries them too, and if it fails, the process exits
 non-zero.
 #### Report limits
 `POST /api/v1/reports` takes reports from anyone who can reach it, without
 credentials, so it is bounded instead. Each refusal below answers with the same
 `{"status":"error"}` body as any other error.
 - **JSON only.** A report must be sent with `Content-Type: application/json`,
  parameters such as `charset` allowed, as the page sends it; any other is
  refused with 415, and nothing of it is stored. A page on another origin can
  send a `text/plain` POST, or a form, without a CORS preflight; refusing them
  keeps any site from making its visitors' browsers store reports (see
  [CORS](#cors)).
 - **Rate limit.** Each client address, resolved through `TRUSTED_PROXIES`, may
  send `REPORTS_PER_MINUTE` reports a minute; past that it gets 429 with
  `Retry-After: 60`. The minute slides: reports from the minute before still
  count, fading out over the current one, so an address is sure never to be
  refused only while it sends at most half of `REPORTS_PER_MINUTE` in any 60
  seconds. The page sends one report a minute from each open tab, so the default
  of 60 refuses nothing from up to 30 tabs behind one address, such as a
  household or an office sharing it, however their reports bunch up. Report
  responses also carry `X-RateLimit-Limit`, `X-RateLimit-Remaining` and
  `X-RateLimit-Reset` headers.
 - **Size cap.** The report files in `DATA_DIR` may total at most
  `DATA_DIR_MAX_BYTES`, counting the files already there at start. Reports
  waiting in memory count at their uncompressed size until they are written,
  those of a failed write included; the part of its file written is removed.
  When a report would take the total past the cap, the oldest report files are
  deleted to make room, and each deletion is logged with the file's name and
  size; a file still being written is never deleted. A report is refused with
  507, and nothing of it is stored, only when the reports waiting to be written
  fill the cap on their own, and then no file is deleted. At start, report files
  past the cap, as after lowering it, are deleted the same way. So the cap is
  how much of the newest reports is kept: the default of 1 GiB is small enough
  for any host; set it to the space you can give `DATA_DIR`.
 #### CORS
 The page calls the API from the origin it is served from, so by default the
 server sends no CORS headers, and browsers let no other origin's pages call it.
 To serve the page from elsewhere, list that origin in `CORS_ALLOWED_ORIGINS`
 (for example `https://netwatch.example.com`); pages from a listed origin may
 `GET` and `POST` with a `Content-Type` header. Each entry must be a plain
 origin, `scheme://host` with an optional `:port`, as browsers send it: no path,
 not even a trailing `/`, and no `*`. Any other entry stops the server from
 starting, with an error naming `CORS_ALLOWED_ORIGINS`.
 #### Metrics
 With both `METRICS_USERNAME` and `METRICS_PASSWORD` set, the server serves
 Prometheus metrics at `GET /metrics` to requests with those as their basic auth
 credentials, and answers any other with 401. For each request that reaches the
 health check or `POST /api/v1/reports`, those the rate limit refuses included,
 the metrics record its duration and response size, labelled with its path,
 method and status; they also count those requests in progress, and include Go's
 runtime and process metrics. No other request is recorded: not those to
 `/metrics` itself, and not those answered before they reach either route, such
 as a CORS preflight, or a request refused with 404 for a path no route has, 405
 for a method its route does not take, or 413 for declaring a body length over
 the 1 MiB limit. A report whose body goes over the limit without declaring its
 length reaches the route, is answered 413 there, and is recorded with that
 status. Clients can make up any number of paths and methods, and each would add
 labels to the metrics for as long as the server runs.
 `/metrics` is rate limited, so that its password cannot be guessed quickly: each
 client address, resolved through `TRUSTED_PROXIES`, may make 60 requests to it a
 minute, whatever their credentials. Past that it gets 429 with
 `Retry-After: 60`, and its credentials are not checked. The minute slides as it
 does for reports (see [Report limits](#report-limits)), so a scraper polling
 every 2 seconds or less often is never refused. This allowance is apart from the
 one for reports.
 #### Sentry
 With `SENTRY_DSN` set, each panic in a handler is reported to that Sentry
 project under the release `netwatch-server-` followed by the server's version.
 On shutdown the server waits up to 2 seconds for Sentry to finish sending.
 ### Monitoring targets
- **26 WAN hosts**: datavi.be (pinned at start), Anthropic API, OpenAI API, AWS
+- **22 WAN hosts**: datavi.be, Anthropic API, OpenAI API, AWS Console, GCP
-  Console, Google Cloud Console, Microsoft Azure, Cloudflare, Fastly CDN,
+  Console, Azure, Cloudflare, Fastly, Akamai, GitHub, B2, 7 S3 regional
-  Akamai, Google, GitHub, B2, 8 S3 regional endpoints (Cape Town, London,
+  endpoints (Cape Town, London, Bahrain, Tokyo, Sydney, Oregon, São Paulo), 4
-  Bahrain, Tokyo, Singapore, Sydney, Oregon, São Paulo) and 6 Hetzner speed test
+  GCS locational endpoints (Iowa, Belgium, Singapore, Sydney)
  servers (Nuremberg, Falkenstein, Helsinki, Ashburn, Hillsboro, Singapore)
 - **Local CPE**: Cable modem at 192.168.100.1 (always monitored)
 - **Local Gateway**: Auto-detected on startup by probing common default gateway
  addresses (192.168.1.1, 192.168.0.1, 192.168.8.1, 10.0.0.1); first responder
@@ -334,28 +159,15 @@ Local hosts are tracked separately from WAN stats.
 ### Latency measurement
-GET requests to each target's URL as written, with `mode: 'no-cors'` and
+HEAD requests with `mode: 'no-cors'` and `cache: 'no-store'`, timed with
-`cache: 'no-store'`, timed with `performance.now()`. No query string is added:
+`performance.now()`. Each check times out after 80% of the refresh interval (24
-the Hetzner speed-test servers close the connection without an answer when the
+seconds at 30 seconds) and is then recorded as a timeout, so a round's checks
-URL has one, and `no-store` keeps the browser's cache out of the measurement.
+have all finished before the next round is due. When no WAN host answers, a
-Each check times out after 80% of the refresh interval (24 seconds at 30
+recovery probe checks 4 random WAN hosts every half second, giving up the checks
-seconds) and is then recorded as a timeout, so a round's checks have all
+it started half a second before. As soon as one answers, a new round starts at
-finished before the next round is due. When no WAN host answers, a recovery
+once, as it does after an interval change. A round started early gives up the
-probe checks 4 WAN hosts, picked at random when it starts, every half second,
+last round's checks if they are still waiting, and that round records nothing
-giving up the checks it started half a second before. As soon as one answers, a
+more, so rounds never overlap. IPv4 only.
 new round starts at once, as it does after an interval change. A round started
 early gives up the last round's checks if they are still waiting, and that round
 records nothing more, so rounds never overlap. The browser chooses between IPv4
 and IPv6 for each target, as for any request; the local targets are IPv4
 addresses.
 Each recorded check that fails writes one line to the browser console with
 `console.error`, and the same line to the debug log: the target's name and URL,
 the time, whether it timed out, answered over the time limit or hit a network
 error (with the error the browser gives the page, such as
 `TypeError: Failed to fetch`, which does not say why), and how long the request
 took. A target that answers after failed checks writes one `console.info` line
 with its latency and how many checks in a row had failed.
 ### Color coding
@@ -380,62 +192,39 @@ dist/
 ## Features
- A round of checks every 3 seconds by default; the interval menu sets 1, 2, 3,
+- Real-time monitoring with 2s update interval and 300s history sparklines
-  5, 10, 15, 30 or 60 seconds and clears the history
+- Health indicator: green (HEALTHY) or red (DEGRADED) based on WAN reachability
- Sparklines of each target's last 100 rounds: 300 seconds at 3 seconds
+- Summary stats: reachable count, min/max/avg latency across WAN hosts only
- The first round after loading or an interval change is discarded, as DNS and
+- Fixed chart axes: Y-axis 0–1000ms, X-axis 0–300s
  TLS setup inflate its latencies
 - Health indicator from the WAN hosts' latest results: OFFLINE (red) when more
  than 10 fail and at most 4 answer, otherwise DEGRADED (orange) when more than
  4 fail, otherwise SLOW (yellow) when more than 3 take over 1000ms, otherwise
  HEALTHY (green)
 - Summary stats across WAN hosts only: how many answered, the min, median,
  average and max of their latest latencies, the min and max over the whole
  history, and the number of rounds run (`Checks`)
 - Fixed chart axes: Y-axis 0–1000ms, higher latencies drawn at the top; X-axis
  the time the history spans
 - Color-coded latency figures and sparkline line segments
 - WAN host rows sorted by latest latency, unreachable last; pinned rows stay on
  top, in name order
 - Play/pause: pause stops probes but history keeps scrolling (blank gaps, no
  false outage)
 - Debug log panel, behind a checkbox in the footer, with five levels (error,
  warning, notice, info, debug) and the last 1000 lines
 - Local and UTC clocks
 - Clickable service URLs
 - A footer link to the commit the page was built from
 - Canvas-based sparkline rendering with devicePixelRatio scaling
 - Zero runtime dependencies: all resources bundled into build artifacts
 ## Deployment
-`make build` writes the page to `dist/`, which any static file host (S3, GCS,
+After running `yarn build`, deploy the contents of the `dist/` directory to any
-Cloudflare Pages, Vercel, Netlify, GitHub Pages) can serve; with no backend
+static file host (S3, GCS, Cloudflare Pages, Vercel, Netlify, GitHub Pages) or
-there, its reports fail quietly and nothing is stored. Or run the Docker image
+use the Docker image behind a reverse proxy.
 behind a reverse proxy.
-The Docker image, built from `Dockerfile` by `make docker`, is the whole service
+The Docker image, built from `Dockerfile`, is the whole service in one
-in one container: nginx serves the built frontend and passes `/api/`,
+container: nginx serves the built frontend and passes `/api/`,
-`/.well-known/healthcheck` and `/metrics` to the Go backend, `netwatch-server`.
+`/.well-known/healthcheck` and `/metrics` to the Go backend, `netwatch-server`,
-`bin/entrypoint.sh` starts it listening only inside the container, on
+which listens only inside the container, on `127.0.0.1:8081`. The image:
 `127.0.0.1:8081`, with `TRUSTED_PROXIES=127.0.0.1/32`, so it takes the client
 address nginx passes on and no other. The image:
 - Listens on port 8080 by default (override with `PORT` env var)
 - Takes the client address from `X-Forwarded-For` only on requests from the
-  reverse proxies named in `TRUSTED_PROXIES`, by default from any address in
+  reverse proxies named in `TRUSTED_PROXIES`, and by default from none
  `10.0.0.0/8`, `172.16.0.0/12` or `192.168.0.0/16`. `bin/entrypoint.sh` checks
  each entry with `netwatch-server check-cidr`, which parses it as the backend
  parses its own `TRUSTED_PROXIES`
 - Sends access logs to stdout
 - Caches static assets with immutable headers
 - Sends the security headers `REPO_POLICIES.md` requires on every response, as
-  `configs/security-headers.conf` sets them, in place of the backend's own
+  `security-headers.conf` sets them, in place of the backend's own
 - Stores reports in `DATA_DIR`, `/data/reports` by default, on the `/data`
  volume. Before the backend starts, the image creates `DATA_DIR` and gives it
-  and `/data` to user `netwatch` (uid 1000), which the backend runs as, with
+  and `/data` to user `netwatch` (uid 1000), which the backend runs as, so a
-  `netwatch-server prepare-data-dir`, which acts on nothing outside `/data`, so
+  host directory bind-mounted at `/data` ends up owned by uid 1000
  a host directory bind-mounted at `/data` ends up owned by uid 1000
 - Writes buffered reports to disk on `docker stop`, and exits non-zero if nginx
  or the backend exits on its own, so the platform restarts it
@@ -445,9 +234,9 @@ What the [upaas](https://git.eeqj.de/sneak/upaas) app for netwatch needs:
 - **Port:** container port `8080`.
 - **Volume:** container path `/data`; the reports are kept in `/data/reports`.
- **Environment variables:** none is required. An empty one counts as unset,
+- **Environment variables:** none is required. An empty one counts as unset, and
-  except `TRUSTED_PROXIES`, and one set to a value netwatch cannot use stops the
+  one set to a value netwatch cannot use stops the container at start, with the
-  container at start, with the reason in its log.
+  reason in its log.
    - `PORT`, default `8080`: the container port, from 1 to 65535. `8081` cannot
      be used: the backend listens on it inside the container
    - `REPORTS_PER_MINUTE`, default `60`: reports each client address may send a
@@ -461,27 +250,22 @@ What the [upaas](https://git.eeqj.de/sneak/upaas) app for netwatch needs:
      in: `/data` or a path below it, with no `.` or `..` part and no extra `/`.
      The container also stops if the path goes through a symbolic link that
      leads out of `/data` or is written as a full path
-    - `TRUSTED_PROXIES`, default `10.0.0.0/8,172.16.0.0/12,192.168.0.0/16`: the
+    - `TRUSTED_PROXIES`, default empty: set it to the address the reverse proxy
-      addresses the reverse proxy in front of the container connects from, as IP
+      in front of the container connects from, as an IP address or CIDR; several
-      addresses or CIDRs separated by commas. nginx takes the client address
+      are separated by commas. nginx takes the client address from
-      from `X-Forwarded-For` only on a request from one of them, and the rate
+      `X-Forwarded-For` only on a request from one of them, and the rate limit
-      limit counts that address. A list replaces the default. Set and empty,
+      counts that address. Unset, `X-Forwarded-For` is ignored and every client
-      `X-Forwarded-For` is ignored and every client behind the proxy shares the
+      behind the proxy shares the proxy's one allowance of `REPORTS_PER_MINUTE`.
-      proxy's one allowance of `REPORTS_PER_MINUTE`. Any client that connects
+      Name only addresses nothing but the proxy connects from: any client that
-      from one of these addresses can write its own `X-Forwarded-For`, and
+      connects from one can write its own `X-Forwarded-For`, and through a port
-      through a port Docker publishes, every client may connect from the Docker
+      Docker publishes, every client may connect from the Docker network's
-      network's gateway, such as `172.17.0.1`, which the default includes; where
+      gateway, such as `172.17.0.1`.
      that matters, name only the addresses nothing but the proxy connects from.
    - `METRICS_USERNAME` and `METRICS_PASSWORD`, default empty: with both set,
      the backend records Prometheus metrics of its requests and serves them at
      `/metrics` on the container port, to requests with this user name and
      password as their basic auth credentials. With neither set, there are no
      metrics and `/metrics` is not found. One set without the other, or a user
      name containing `:`, stops the container
    - `SENTRY_DSN`, default empty: set to a Sentry project's DSN, the backend
      sends its errors to that Sentry project: each request whose handling
      crashes, which still gets a 500 response. A value Sentry does not accept
      stops the container. Empty, the backend sends nothing to Sentry
 - **Health check:** the image's `HEALTHCHECK` requests
  `/.well-known/healthcheck` through nginx every 30 seconds, so it fails unless
  both nginx and the backend answer. upaas reads the container's health 60
@@ -495,19 +279,19 @@ properties.
 ## Limitations
- **CORS**: The checks are cross-origin requests in `no-cors` mode, so the page
+- **CORS**: Some hosts may block cross-origin HEAD requests. The app uses
-  cannot read the answer, only time it: any answer counts as reachable, an error
+  `no-cors` mode which allows the request but provides opaque responses. Latency
-  page included.
+  is still measurable based on request timing.
- **Local targets**: The cable modem at 192.168.100.1 and the detected gateway
+- **Local gateway**: The 192.168.100.1 endpoint requires the host to be
-  answer only on a network that has them, and only when NetWatch is served from
+  accessible from your network.
  localhost or a private address (see Monitoring targets).
 - **Network conditions**: Measurements reflect browser-to-endpoint latency,
  which includes your local network, ISP, and internet routing.
 ## TODO
-The to-do list is [TODO.md](TODO.md): where the work stands, the next step, the
+- Add configurable host list (environment variable or config file)
-open work, and what has been done.
+- Add latency history export (CSV/JSON)
 - Add notification/alert when status changes to DEGRADED
 ## License
@@ -1,6 +1,6 @@
 ---
 title: Repository Policies
-last_modified: 2026-10-04
+last_modified: 2026-07-06
 ---
 This document covers repository structure, tooling, and workflow standards. Code
@@ -60,28 +60,17 @@ style conventions are in separate documents:
  prerequisite since nvm requires bash. yarn is then pinned via
  `corepack prepare yarn@<version> --activate`. Never install "latest" or "lts";
  always exact versions. `script/cibuild` runs the CI build: it changes to the
-  repo root, runs `script/bootstrap`, runs `script/check`, and builds the image
+  repo root and runs `docker build .`; the Gitea workflow calls it. Four further
-  with the version; the Gitea workflow calls it. **`script/cibuild` runs
+  scripts are our own extensions to the standard: `script/check` runs
-  `script/bootstrap` first**, because the workflow checks out the repo and runs
+  `script/test`, `script/lint`, and `script/fmt-check`; `script/precommit` is
-  nothing else, while `script/fmt-check` runs the formatter on the host: on a
+  what the git pre-commit hook runs, and it calls `script/check`;
-  pristine checkout with nothing installed the run dies there, after the
+  `script/install-precommit` installs the git pre-commit hook (the `make hooks`
-  containerised gates have passed. **The bootstrap alone is not enough**:
+  target shims to it); and `script/projectname` (literally that filename) simply
-  `script/bootstrap` installs node and yarn under nvm and leaves neither on the
+  outputs the project's name. Scripts that need the name call
-  `PATH` of the shell that called it, so a bare `yarn` still exits 127. The host
+  `script/projectname` — e.g. `script/docker` assembles its image tag from it —
-  entrypoints that need yarn — `script/fmt` and `script/fmt-check` — therefore
+  so those scripts stay byte-identical across all repos. Repo-type-specific
-  source nvm for the pinned node version before invoking it, exactly as
+  pre-commit extras (e.g. `go mod tidy` verification in Go repos) belong in
-  `script/bootstrap`'s own install step does. A runner carrying nothing but
+  `script/precommit`, not in the hook itself. Model scripts are at
  docker and git then gets through `script/check`. Four further scripts are our
  own extensions to the standard: `script/check` runs `script/test`,
  `script/lint` and `script/fmt-check`; `script/precommit` is what the git
  pre-commit hook runs, and it calls `script/check`; `script/install-precommit`
  installs the git pre-commit hook (the `make hooks` target shims to it); and
  `script/projectname` (literally that filename) simply outputs the project's
  name. Scripts that need the name call `script/projectname` — e.g.
  `script/docker` assembles its image tag from it — so those scripts stay
  byte-identical across all repos. Repo-type-specific pre-commit extras (e.g.
  `go mod tidy` verification in Go repos) belong in `script/precommit`, not in
  the hook itself. Model scripts are at
  `https://git.eeqj.de/sneak/prompts/raw/branch/main/script/<name>`. The README
  must document the provided scripts in an **Entrypoints** section (see the
  README requirements below).
@@ -100,198 +89,87 @@ style conventions are in separate documents:
  contributor should be able to understand the entire development workflow by
  reading the Makefile.
- Every repo should have a `Dockerfile`, and it carries the repo's gates: a
+- Every repo should have a `Dockerfile`. All Dockerfiles must run `make check`
-  `lint` phase and a `test` phase, with the final stage depending on both so the
+  as a build step so the build fails if the branch is not green. For non-server
-  image cannot be built unless they pass. For non-server repos the final stage
+  repos, the Dockerfile should bring up a development environment and run
-  brings up a development environment; for server repos it is the runtime image.
+  `make check`. For server repos, `make check` should run as an early build
-  The gate phases and the build stage start from their pinned base images and
+  stage before the final image is assembled. Dockerfiles install development
-  install what those images lack either inline, as the canonical Go `Dockerfile`
+  prerequisites by running `script/bootstrap` rather than duplicating installs
-  below does for `git`, or by running `script/bootstrap`, as the `prompts`
+  inline; COPY `script/` and the dependency manifests (`package.json` +
-  repo's own `Dockerfile` does for its yarn packages. The development
+  `yarn.lock`, `go.mod` + `go.sum`, etc.) before running it so the bootstrap
-  environment stage installs development prerequisites by running
+  layer stays cached until dependencies change.
  `script/bootstrap` rather than duplicating its installs inline. A stage that
  runs `script/bootstrap` COPYs `script/` and the dependency manifests
  (`package.json` + `yarn.lock`, `go.mod` + `go.sum`, etc.) before running it.
- **Linting and testing run in Docker, as phases of the `Dockerfile`.** There is
+- **Dockerfiles must use a separate lint stage for fail-fast feedback.** Go
-  no separate lint file. `script/lint` and `script/test` each build one phase
+  repos use a multistage build where linting runs in an independent stage based
-  and nothing else:
+  on the `golangci/golangci-lint` image (pinned by hash). This stage runs
  `make fmt-check` and `make lint` before the full build begins. The build stage
  then declares an explicit dependency on the lint stage via
  `COPY --from=lint /src/go.sum /dev/null`, which forces BuildKit to complete
  linting before proceeding to compilation and tests. This ensures lint failures
  surface in seconds rather than minutes, without blocking on dependency
  download or compilation in the build stage.
-    ```sh
+    The standard pattern for a Go repo Dockerfile is:
    docker build --no-cache --target lint -t "$(script/projectname)-lint" .
    docker build --no-cache --target test -t "$(script/projectname)-test" .
    ```
    **A stage that is not the last one in the file is built only when the final
    stage's chain depends on it, or when `--target` names it.** That is why the
    two gates are always invoked by name here, and why the final stage carries a
    `COPY --from=` of a harmless file from each of them: without that edge a
    plain `docker build .` builds the last stage alone and exits 0 having linted
    and tested nothing.
    **Every `docker build` in `script/` is tagged**, here and in
    `script/cibuild` and `script/docker`. An untagged build leaves a dangling
    image behind on every invocation, on every developer host and every CI
    runner; a tagged one replaces the previous image.
    Inside a phase the tool is invoked directly — `golangci-lint`, `go test`,
    `eslint`, `prettier` — never through `make lint` or `script/test`, which are
    themselves a `docker build` and would recurse into a daemon that does not
    exist in a build step. Formatting is the exception and stays on the host:
    `script/fmt` writes the working tree, and `script/fmt-check` is its
    read-only twin.
    **No lint verdict may come from a host invocation of the linter.** On a
    shared host golangci-lint reads a result cache keyed on file content rather
    than location, so a second checkout of the same content is served the first
    one's findings, and a host-global lock in `$TMPDIR` makes concurrent runs
    exit non-zero with `parallel golangci-lint is running` — a status a caller
    cannot tell from real findings. Both have produced wrong verdicts in this
    org, in both directions. A container has its own cache, its own `TMPDIR` and
    a digest-pinned binary, so neither is reachable.
 - **Any build that runs checks is built with `--no-cache`.** Docker invalidates
  a `COPY` layer only when the copied content changes, so on an unchanged tree
  the check `RUN` is served from cache, nothing executes, and the build still
  exits 0. Every `docker build` in `script/` therefore passes `--no-cache`:
  `script/lint`, `script/test`, `script/cibuild` and `script/docker` are the
  four, and there is no fifth — `script/check` runs the two gate phases and
  `script/fmt-check`, and builds no image of its own. A bare `docker build .` is
  not evidence that anything ran: a sub-second build reporting success is a
  cache hit, not a result. Never invalidate by pruning — `docker builder prune`
  and friends destroy a build cache shared with every other build on the host.
  When a check is added or changed, prove it works by planting a defect it must
  catch and watching the run fail on it, then revert the defect. A green run
  alone shows neither that the check ran nor that it covers what it should.
 - **The gate phases are separate stages, and the build stage depends on both.**
  The lint phase is based on the `golangci/golangci-lint` image (pinned by
  hash), so lint failures surface in seconds rather than after a full compile,
  and the test phase is based on the Debian Go image. The canonical Go repo
  `Dockerfile`:
    ```dockerfile
-    # Lint phase
+    # Lint stage — fast feedback on formatting and lint issues
    # golangci/golangci-lint:v2.x.x, YYYY-MM-DD
    FROM golangci/golangci-lint@sha256:... AS lint
    WORKDIR /src
    COPY go.mod go.sum ./
    RUN go mod download
    COPY . .
-    RUN golangci-lint run --config .golangci.yml ./...
+    RUN make fmt-check
    RUN make lint
-    # Test phase. -race needs cgo and so a C compiler, which the Debian Go
+    # Build stage
    # image ships and the alpine one does not.
    # golang:1.x, YYYY-MM-DD
    FROM golang@sha256:... AS test
    WORKDIR /src
    COPY go.mod go.sum ./
    RUN go mod download
    COPY . .
    RUN go test -timeout 90s -race -cover ./... || \
        { echo "--- Rerunning with -v for details ---"; \
          go test -timeout 90s -race -v ./...; exit 1; }
    # Build stage. Nothing is wanted from either phase above; the copies
    # are what make BuildKit build them first, so this stage cannot run
    # unless lint and test passed.
    # golang:1.x-alpine, YYYY-MM-DD
    FROM golang@sha256:... AS builder
    COPY --from=lint /src/go.sum /dev/null
    COPY --from=test /src/go.sum /dev/null
    RUN apk add --no-cache git
    # A tar-stream context keeps the sender's file owners, which git refuses.
    RUN git config --system --add safe.directory /src
    WORKDIR /src
    # Force BuildKit to run the lint stage before proceeding
    COPY --from=lint /src/go.sum /dev/null
    COPY go.mod go.sum ./
    RUN go mod download
    COPY . .
    RUN make test
-    # The VERSION build arg when one is given, otherwise
+    ARG VERSION=dev
-    # `git describe --tags --always` on the .git in the build context. With
+    RUN CGO_ENABLED=0 go build -trimpath \
    # .git present, a version that is still empty, dev or unknown fails the
    # build: git is missing or could not read the checkout.
    ARG VERSION
    RUN VERSION="${VERSION:-$(git describe --tags --always)}"; \
        if [ -e .git ]; then \
            case "$VERSION" in ""|dev|unknown) \
                echo "version is '$VERSION' although .git is present" >&2; \
                exit 1 ;; \
            esac; \
        fi; \
        CGO_ENABLED=0 go build -trimpath \
        -ldflags="-s -w -X main.Version=${VERSION}" \
        -o /app ./cmd/app/
-    # Runtime stage, and the last one
+    # Runtime stage
    FROM alpine@sha256:...
    COPY --from=builder /app /usr/local/bin/app
    ENTRYPOINT ["app"]
    ```
    Key points:
-    - The lint phase uses the `golangci/golangci-lint` image directly (it has
+    - The lint stage uses the `golangci/golangci-lint` image directly (it
-      both Go and the linter), so nothing needs installing.
+      includes both Go and the linter), so there is no need to install the
-    - `COPY --from=<phase> /src/go.sum /dev/null` is a no-op copy whose only
+      linter separately.
-      purpose is the ordering edge. BuildKit runs stages in parallel by default,
+    - `COPY --from=lint /src/go.sum /dev/null` is a no-op file copy that creates
-      and a stage nothing depends on is not built at all, so without these two
+      a stage dependency. BuildKit runs stages in parallel by default; without
-      lines a red gate would not fail the build.
+      this line, the build stage would not wait for lint to finish and a lint
-    - Keep the runtime stage last, and if you add a stage after it, give it the
+      failure might not fail the overall build.
      same two copies. A plain `docker build .` builds the last stage's chain
      and nothing else.
    - If the project uses `//go:embed` directives that reference build artifacts
-      (e.g. a web frontend compiled in a separate stage), the lint phase must
+      (e.g. a web frontend compiled in a separate stage), the lint stage must
      create placeholder files so the embed directives resolve. Example:
      `RUN mkdir -p web/dist && touch web/dist/index.html web/dist/style.css`.
-    - If the project requires CGO or system libraries for linting, install them
+      The lint stage should not depend on the actual build output — it exists to
-      in the lint phase. The `golangci/golangci-lint` image is Debian-based and
+      fail fast.
-      has no `apk`, so install with `apt-get` under the Debian package name
+    - If the project requires CGO or system libraries for linting (e.g.
-      (`libvips-dev`, where alpine says `vips-dev`), and delete the package
+      `vips-dev`), install them in the lint stage with `apk add`.
-      lists in the same `RUN`, so the layer does not keep them:
+    - The build stage runs `make test` after compilation setup. Tests run in the
-
+      build stage, not the lint stage, because they may require compiled
-        ```dockerfile
+      artifacts or heavier dependencies.
        RUN apt-get update \
            && apt-get install -y --no-install-recommends libvips-dev \
            && rm -rf /var/lib/apt/lists/*
        ```
    - `.dockerignore` lets `.git` into the build context. It keeps out every git
      `config` at any depth (`**/.git/config`, `**/.git/modules/**/config`): the
      repository's own, each submodule's under `.git/modules/`, and that of a
      submodule keeping its own `.git` directory. `git describe` does not need
      them, and each can hold a credential: a password in a remote URL, or the
      token the CI checkout step stores there. A submodule whose name has a
      `config` segment (`config`, `deploy/config`, `config/lib`) loses its whole
      git directory to `**/.git/modules/**/config`, and Go's version stamping
      then fails the build: give it a name without that segment
      (`git submodule add --name`). The stage that compiles has `git` (the
      Debian Go image has it; an alpine one needs `apk add --no-cache git`) and
      takes the version from the `VERSION` build argument when one is given,
      otherwise from `git describe --tags --always`. That gives the tag on a
      tagged commit; on a later commit, the tag, the number of commits since it
      and the short commit (`v1.2.3-4-gabc1234`); and the short commit when no
      tag is reachable. The stage that compiles also marks its working directory
      safe for git (`git config --system --add safe.directory /src`): a context
      sent as a tar stream keeps the sender's file owners, and git refuses a
      checkout owned by another user, so the version would come out empty.
      `ARG VERSION` has no default, and the build fails if the context carries
      `.git` and the version still comes out empty, `dev` or `unknown`. A plain
      `docker build .` with no build arguments must succeed; a Dockerfile that
      refuses an empty build argument drops that refusal and keeps the argument.
 - Every repo should have a Gitea Actions workflow (`.gitea/workflows/`) that
-  runs `script/cibuild` on push, and checks out the repo as its only other step.
+  runs `script/cibuild` (which runs `docker build .`) on push. Since the
-  That script bootstraps, runs the gate phases, and then builds the image, so a
+  Dockerfile already runs `make check`, a successful build implies all checks
-  successful run means every check passed; a bare `docker build .` does not
+  pass.
  carry the same guarantee, because its gate phases may come from the cache. The
  image build is uncached and so runs the gate phases a second time. That is the
  price of the rule above, and it is worth paying: the image that ships is built
  from a run of its own gates rather than from a cache entry. A separate
  workflow limited to `main` by a `branches` list under `on: push` cannot be
  checked by review: to try a change to it, add the feature branch to that list
  and push, then remove the branch from the list again before merging. Keep any
  job in it that publishes behind `if: github.ref_name == 'main'`, so the run
  from the feature branch publishes nothing.
 - Use platform-standard formatters: `black` for Python, `prettier` for
  JS/CSS/Markdown/HTML, `go fmt` for Go. Always use default configuration with
@@ -311,21 +189,14 @@ style conventions are in separate documents:
  module under test to verify it compiles/parses. There is no excuse for
  `make test` to be a no-op.
- `make test` must complete in under 60 seconds. That is the hard cap, and a
+- `make test` must complete in under 20 seconds. Add a 30-second timeout in the
-  suite that exceeds it fails. Under 20 seconds is the target. A suite between
+  Makefile.
  20 and 60 seconds is still green, but the overage must be filed as an
  improvement bug against that repo. Add a 90-second timeout to the test
  invocation (`go test -timeout 90s`). The backstop deliberately sits above the
  hard cap so that it catches a genuinely hung test rather than a merely slow
  one.
- **The test command should use the conditional verbose rerun pattern.** Run
+- **`make test` should use the conditional verbose rerun pattern.** Run tests
-  tests without `-v` (verbose) first. If tests fail, automatically rerun with
+  without `-v` (verbose) first. If tests fail, automatically rerun with `-v` to
-  `-v` to show full output. This keeps CI logs and `docker build` output clean
+  show full output. This keeps CI logs and `docker build` output clean on
-  on success (just package/suite summaries) while providing full diagnostic
+  success (just package/suite summaries) while providing full diagnostic detail
-  detail on failure (every test case, every assertion). The command lives in the
+  on failure (every test case, every assertion). The general shell pattern:
  `test` phase of the `Dockerfile`, since `script/test` builds that phase; the
  Makefile form below is the same pattern for any repo-local invocation:
    ```makefile
    test:
@@ -338,26 +209,11 @@ style conventions are in separate documents:
    ```makefile
    test:
-    	@go test -count=1 -timeout 90s -race -cover ./... || \
+    	@go test -timeout 30s -race -cover ./... || \
    		{ echo "--- Rerunning with -v for details ---"; \
-    		  go test -count=1 -timeout 90s -race -v ./...; exit 1; }
+    		  go test -timeout 30s -race -v ./...; exit 1; }
    ```
    `-count=1` is required on both invocations: it defeats Go's test _result_
    cache, so neither run can report a stored pass in place of running the
    tests. It leaves the build cache alone, so it costs the runtime of the suite
    and no recompilation.
    That cache is Go's own, separate from Docker's layer cache. Go stores a
    passing result in its cache directory (`GOCACHE`), and when the same tests
    run again on unchanged code it prints that result, marked `(cached)`,
    without running them. That matters on a developer's machine, where this
    target runs and the directory lasts from one run to the next. The `test`
    phase of the `Dockerfile` needs no `-count=1`: its base image holds no
    result for this repo's tests and nothing before its `go test` step runs a
    test, so there is nothing to replay. `--no-cache` (above) is what makes that
    step run on an unchanged tree.
    Python example:
    ```makefile
@@ -383,84 +239,10 @@ style conventions are in separate documents:
  must be in `.gitignore`. No exceptions.
 - `.gitignore` should be comprehensive from the start: OS files (`.DS_Store`),
-  editor files (`.swp`, `*~`), in-repo agent scratch directories (`.claude/`),
+  editor files (`.swp`, `*~`), language build artifacts, and `node_modules/`.
-  language build artifacts, and `node_modules/`. Fetch the standard `.gitignore`
+  Fetch the standard `.gitignore` from
-  from `https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when
+  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when setting up
-  setting up a new repo. These patterns are written to `.gitignore`'s own
+  a new repo.
  semantics, in which an unanchored pattern already matches at every depth; they
  are not a `.dockerignore` and must not be transplanted into one unmodified.
 - **`.dockerignore` does not use `.gitignore` semantics, and copying patterns
  across unmodified leaves secrets in the build context.** Docker matches with
  `moby/patternmatcher`: `filepath.Match` semantics plus a `**` extension, so
  `*` does not cross `/` and a pattern without a leading `**/` is anchored at
  the build-context root. A `.dockerignore` listing `.env`, `*.pem` and `*.key`
  therefore excludes only the copies at the repository root, while `config/.env`
  and `certs/server.key` still reach the context and can land in an image layer
  — which is more dangerous than a short file with no secret patterns at all,
  because it reads as solved and stops anyone looking. Give every
  depth-independent pattern the `**/` prefix and leave only genuinely
  root-anchored entries unprefixed: `.claude`, and the repo's own host-built
  binary, written `/myapp` and never `**/myapp`, which would also match
  `cmd/myapp/` and delete the package directory from the context. Matching is
  case-sensitive, and an ALL-CAPS twin per pattern still misses `Server.Key`, so
  secret names use character ranges — `**/*.[kK][eE][yY]`, `**/*.[pP][eE][mM]`,
  and likewise for `.envrc` and the extensionless SSH keys. Where such a pattern
  also catches something the build needs, re-include it with a negation
  (`!docs/example.env`); deleting the pattern reopens the exposure for every
  other file it covers. Fetch the standard `.dockerignore` from
  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.dockerignore` and extend
  it with the repo's own artifacts.
 - **In-repo agent scratch belongs in both files, written to each file's own
  semantics.** `.claude/` holds one worktree per in-flight agent — an entire
  additional checkout of the repo — so under `COPY . .` the build context
  inflates by a multiple of the repo and another session's unreviewed work can
  be copied into an image layer. In `.gitignore` the entry is `.claude/`,
  unanchored. In `.dockerignore` it is `.claude`, anchored and with **no** `**/`
  prefix, because the prefixed form would also delete any nested directory of
  that name from the build. Anchoring carries a known gap that the canonical
  `.dockerignore` states in its own comment, since consuming repos receive the
  file and not the tracker: the directory is created in the agent's working
  directory, so a repo running agents in subdirectories still ships
  `services/api/.claude/` and must add its own anchored entry there.
 - **A plain `docker build .` of a clone stamps the version that
  `git describe --tags --always` gives**, derived from the `.git` in the build
  context as the canonical `Dockerfile` above shows. Without its failure check,
  a missing `git` or an unreadable checkout would leave `-X main.Version=` empty
  and the build would still exit 0. `script/docker` and `script/cibuild` pass
  the version they compute on the host; it takes precedence. They do this
  byte-identically across repos:
    ```sh
    # Own line: a failing command substitution inside an argument does not
    # trip `set -e`, so the inline form degrades to an empty constant.
    version="$(git describe --tags --always --dirty 2>/dev/null || true)"
    [ -n "$version" ] || version="unknown"
    docker build --no-cache \
        --build-arg VERSION="$version" \
        -t "$(script/projectname)" .
    ```
    `--always` makes an untagged repo yield an abbreviated commit hash rather
    than failing, and the `[ -n "$version" ]` line is the single place the
    fallback is applied — a live check that fires on a build from an export with
    no `.git` and on a repository with no commits yet. Do not fold it into the
    substitution as `|| echo unknown`, which makes the guard unreachable. The
    Dockerfile's side is `ARG VERSION` in the stage that compiles, declared
    there because `ARG` is stage-scoped; passing `VERSION` to a repo whose
    Dockerfile declares no such `ARG` is ignored and costs nothing, which is why
    the scripts stay byte-identical. One consequence for CI: the standard
    checkout action clones shallow and fetches no tags, so a repo that embeds a
    tag-derived version must set `fetch-depth: 0` on its checkout step.
 - **Verify `.dockerignore` by enumerating the image, not by reading the
  patterns.** Plant files at the root _and_ at least two directories deep, build
  a probe image that does `COPY . .`, and list what actually landed
  (`docker run --rm --entrypoint find IMAGE /app`). The `transferring context`
  size is not a substitute: a nested secret is a few bytes, and BuildKit
  transfers only the delta from the previous build.
 - **No build artifacts in version control.** Code-derived data (compiled
  bundles, minified output, generated assets) must never be committed to the
@@ -476,56 +258,9 @@ style conventions are in separate documents:
 - Make all changes on a feature branch. You can do whatever you want on a
  feature branch.
- `.golangci.yml` is standardized. The vendored copy in a consuming repo must
+- `.golangci.yml` is standardized and must _NEVER_ be modified by an agent, only
-  _NEVER_ be modified by an agent: fetch it from
+  manually by the user. Fetch from
-  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.golangci.yml` and keep it
+  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.golangci.yml`.
  byte-identical, so that no repo can quietly loosen its own linting. Linter
  configuration changes are made to the canonical copy in the `prompts` repo and
  reach consuming repos by re-vendoring; an agent may open a PR against
  canonical, which only the user merges. One list is exempt from byte-identity,
  because it cannot be written once for every repo: the `deny` list of the
  `test-support` depguard rule, where a repo names its own test-support packages
  by full import path. A repo adds entries there and changes nothing else, and a
  re-vendor carries its entries forward. The canonical golangci-lint version is
  v2.14.0 (released 2026-09-24), pinned as the digest of the lint phase's base
  image
  (`golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a92406b2fd3898faf0bf98f`,
  which reports `2.14.0 built with go1.27.0 from 114493f9`). A module's `go`
  directive must not name a newer Go minor version than the one golangci-lint
  was built with, or golangci-lint refuses to lint it: this release lints
  `go 1.27.1` but not `go 1.28`. That digest is the only pin, since no repo
  installs golangci-lint on the host. A repo sets the lint phase digest to the
  one named here and re-vendors `.golangci.yml` in the same commit, whichever of
  the two prompted the change: the canonical copy can name linters that an older
  golangci-lint rejects, and a newer golangci-lint can add linters that
  `default: all` switches on until the canonical copy disables them.
 - **`script/bootstrap` installs a pinned tool by comparing versions, never by
  testing presence.** An `if ! command -v <tool>; then install; fi` guard tests
  `PATH` only, so on an already-provisioned machine the pin is inert and a
  version bump is a silent no-op — while the Dockerfile, installing into a clean
  image, gets the pinned version, so a local `make check` and `make docker` can
  disagree about what the tool even is. The canonical form:
    - compares the installed version against the pin over the **whole** version
      token; a parser that stops at the first `-` reports `2.12.2` for a host
      running `2.12.2-rc1` and skips the install;
    - treats absent, non-zero, empty or unrecognised `--version` output as a
      mismatch, so the failure direction is a redundant install and never a
      skipped one;
    - after installing, re-resolves the binary the way callers do — `hash -r`,
      then through `PATH`, not through the directory the installer wrote to —
      and fails naming the resolved path, since an install that a shadowing
      binary hides succeeds while changing nothing any caller sees;
    - is actually called, and prints the version on both success paths: a
      function defined and never invoked has the same exit status and the same
      empty output as one that worked.
    Keep it POSIX sh: no arrays, no `[[`, no `grep -P`.
    A Go tool a repo needs on the host is installed with `go install` pinned to
    a commit hash (`go install <package>@<commit hash>`). It is never tracked as
    a `go.mod` tool dependency or through a `tools.go` file, either of which
    pulls the tool's own dependencies into the repo's `go.mod` and `go.sum`.
 - When pinning images or packages by hash, add a comment above the reference
  with the version and date (YYYY-MM-DD).
@@ -639,14 +374,12 @@ style conventions are in separate documents:
  settings.
 - Avoid putting files in the repo root unless necessary. Root should contain
-  only project-level config files (`README.md`, `AGENTS.md`, `Makefile`,
+  only project-level config files (`README.md`, `Makefile`, `Dockerfile`,
-  `Dockerfile`, `LICENSE`, `.gitignore`, `.editorconfig`, `REPO_POLICIES.md`,
+  `LICENSE`, `.gitignore`, `.editorconfig`, `REPO_POLICIES.md`, and
-  and language-specific config). Everything else goes in a subdirectory.
+  language-specific config). Everything else goes in a subdirectory. Canonical
-  Canonical subdirectory names:
+  subdirectory names:
    - `bin/` — executable scripts and tools
-    - `cmd/` — Go command entrypoints; thin only: one `main.go` per binary whose
+    - `cmd/` — Go command entrypoints
      body is a single call into `internal/` or `pkg/`, no project logic in
      `cmd/`
    - `configs/` — configuration templates and examples
    - `deploy/` — deployment manifests (k8s, compose, terraform)
    - `docs/` — documentation and markdown (README.md stays in root)
@@ -673,7 +406,3 @@ style conventions are in separate documents:
    - Go: `go.mod`, `go.sum`, `.golangci.yml`
    - JS: `package.json`, `yarn.lock`, `.prettierrc`, `.prettierignore`
    - Python: `pyproject.toml`
 - Guidance for coding agents lives in one `AGENTS.md` at the repository root. It
  is never committed under a file or directory named after one agent tool, such
  as `CLAUDE.md` or `.claude/`, and never split into separate memory files.
@@ -1,193 +1,28 @@
 # Workflow
- branch from `next`
+- branch (from `main`)
 - do the work in Next Step
 - move Next Step to the top of Completed Steps
 - move the top item of Future Steps into Next Step
 - commit (`TODO.md` changes in the same commit as the work)
- push the branch and open a PR against `next`
+- merge to `main` if the branch is not protected, otherwise open a PR
 - push
 # Status
-pre-1.0. No git tags. `main` is the stable branch and `next` the development
+pre-1.0. No git tags. `feat/reportbuf-storage` is merged; the backend, the CI
-branch, which every PR targets. The frontend and the Go backend ship as one
+workflow, and the backend repo standard files are all on `main`. Frontend and
-Docker image, and the Gitea workflow `.gitea/workflows/check.yml` runs
+backend are both functional. Working toward the 1.0.0 milestone by closing the
-`script/cibuild` on every push. Working toward 1.0.0.
+remaining repo-compliance issues on the tracker.
 # Next Step
-Run `make frontend-viewport-test` in CI as its own step; it is not part of
+Confirm the `.gitea/workflows/check.yml` run is green (main always green
-`make check`, as it needs Docker and takes minutes.
+policy). The workflow file is already on `main`; what is unverified is that its
 latest run passes.
 # Completed Steps
 - 2026-10-14: stopping the server waits for a report file write already under
  way, the periodic flush's or the one `Append` starts when 10 MiB are waiting,
  before the final flush
  ([#122](https://git.eeqj.de/sneak/netwatch/issues/122)). It used to return at
  once, and the process exiting cut the write off, losing its reports. A write
  under way that fails puts its reports back in memory, so the final flush
  writes them; the stop fails, and the process exits non-zero, only if that
  flush fails too. Two tests hold the periodic write and the one at 10 MiB open
  and find the stop waiting until each is released; a third makes the held write
  fail and finds its report written by the final flush. They run in a
  `testing/synctest` bubble, so the periodic flush comes without waiting a
  minute
 - 2026-10-14: `make test` no longer runs the frontend's production build, which
  the tests do not need ([#118](https://git.eeqj.de/sneak/netwatch/issues/118)).
  `yarn build` moved out of the stage that runs the frontend's unit tests, now
  named `frontend-test`, into a new `frontend-build` stage, which starts from it
  and, like the `builder` stage, waits on the lint and test phases, so the image
  still cannot be built unless they pass. The runtime stage copies `dist/` from
  it, and the `git` that `vite.config.js` reads the commit with is installed
  there. No build cache was added
 - 2026-10-14: opening the debug log no longer raises the measured latencies
  ([#131](https://git.eeqj.de/sneak/netwatch/issues/131)). With the panel open,
  each line written drew the panel again with every line before it, up to 1000,
  and a check that ends while the page draws is timed only when the drawing is
  done. Each line is now added at the end of the open panel, its first line is
  dropped once it holds more than 1000, and it scrolls to its newest line once a
  frame rather than once a line; the whole panel is drawn only when it is
  opened. A unit test writes more lines than the panel keeps and checks that the
  lines already shown are left as they are
 - 2026-10-14: unset, the backend's own `TRUSTED_PROXIES` trusts `10.0.0.0/8`,
  `172.16.0.0/12` and `192.168.0.0/16`, and no longer loopback; a list replaces
  that; set and empty, it trusts no proxy, where it used to get the default, as
  viper reads an empty environment variable as unset
  ([#129](https://git.eeqj.de/sneak/netwatch/issues/129)). A value it cannot
  parse still stops the start, and the image still starts the backend with
  `TRUSTED_PROXIES=127.0.0.1/32`. Tests load the config with it unset, empty and
  set to a list
 - 2026-10-14: while report files cannot be written, `POST /api/v1/reports`
  answers 500, and storing recovers once writing works again, with no restart
  ([#123](https://git.eeqj.de/sneak/netwatch/issues/123)). A failed write puts
  its reports back in the report buffer, still counted against
  `DATA_DIR_MAX_BYTES`, where they used to be lost; the next flush, within a
  minute, tries them again. Until a write succeeds, every report is refused and
  nothing of it is kept, so the buffer does not grow while writes fail. A test
  posts reports through the report handler before, during and after a failure;
  others check that a report appended while a write runs is kept, behind that
  write's own, when the write fails, and that the cap still holds exactly once a
  failed write's reports are written
 - 2026-10-14: a request whose handler panics is logged with status 500 and
  counted with `code` 500 in the metrics
  ([#125](https://git.eeqj.de/sneak/netwatch/issues/125)), not 200. `Recoverer`
  is still the first middleware; `Write500OnPanic`, after the request log and
  after the metrics, writes the 500 as the panic passes and lets it go on to
  Sentry and `Recoverer`, which now writes a 500 only when no status was written
 - 2026-10-14: with `TRUSTED_PROXIES` unset, the container's nginx trusts the
  three private ranges of RFC 1918, `10.0.0.0/8`, `172.16.0.0/12` and
  `192.168.0.0/16` ([#119](https://git.eeqj.de/sneak/netwatch/issues/119)), as
  sneak ruled on 2026-09-22 for every app; it trusted no proxy before. A list
  replaces them, and set and empty it trusts none. The backend inside the
  container still trusts only `127.0.0.1/32`
 - 2026-10-14: `POST /api/v1/reports` takes only a report sent with
  `Content-Type: application/json`, parameters such as `charset` allowed, and
  refuses any other with 415 and the usual error body, storing nothing
  ([#126](https://git.eeqj.de/sneak/netwatch/issues/126)). A page on another
  origin can send a `text/plain` POST without a CORS preflight, so any site
  could make its visitors' browsers store reports; a JSON POST waits for the
  preflight, which only an origin in `CORS_ALLOWED_ORIGINS` passes. The page
  already sent `application/json`. A test posts a report as `text/plain`, as
  `application/x-www-form-urlencoded`, as `multipart/form-data`, with no
  `Content-Type`, and as `application/json` with and without a charset, and
  checks which are stored
 - 2026-10-14: changing the interval no longer drops the samples not yet reported
  to the backend ([#124](https://git.eeqj.de/sneak/netwatch/issues/124)). The
  interval menu resets each host with the new `HostState.reset`, which moves the
  history's entries to `clearedHistory`, keeping the last 100 of all it has
  cleared, and `buildReport` reads both, so the next report has the samples
  recorded before the change that no report has delivered. Until its next check,
  such a host is sent with the status of its last check, not `pending`. The
  page's history still starts over, and the per-host marks still leave out of a
  report the samples a delivered one had. Unit tests reset a host as the
  interval menu does and find its unreported samples in the next report, and
  only the last 100 after more were cleared; `web/src/main.js` now exports
  `Reporter` for them
 - 2026-10-14: the repo has the layout `REPO_POLICIES.md` gives
  ([#30](https://git.eeqj.de/sneak/netwatch/issues/30)). The Go module is at the
  root (`go.mod`, `go.sum`, `.golangci.yml`, `cmd/netwatch-server/`,
  `internal/`), so its path `sneak.berlin/go/netwatch` names the repository.
  `go.mod` ignores `node_modules/`, where npm packages can ship `.go` files that
  `make tidy` would otherwise read as this module's packages. The page's source
  is in `web/` (`index.html`, `src/`), which `vite.config.js` makes Vite's root;
  the build still goes to `dist/`. `nginx.conf` and `security-headers.conf` are
  in `configs/`. `backend/` is gone: its `LICENSE` and `.dockerignore` are
  deleted, its `.gitignore` entries are in the root one, its `README.md` is
  folded into the root `README.md`, and its scripts into the root `script/`:
  `script/build` builds the server binary after the frontend, `script/run` and
  `script/clean` came from `backend/script/`, and `make test`, `make lint`,
  `make fmt` and `make fmt-check` do what its `test`, `lint`, `fmt` and
  `fmt-check` did. `script/bootstrap` no longer installs gcc, which only the
  host Go tests needed. The image and the page are unchanged
 - 2026-10-07: the files shared from `sneak/prompts` are its copies at commit
  `dd4027b` ([#113](https://git.eeqj.de/sneak/netwatch/issues/113)), with this
  repository's own entries after the shared content in `.gitignore`,
  `.dockerignore` and `.editorconfig`. `make lint` and `make test` each build
  one phase of `Dockerfile` without the cache: `lint` runs golangci-lint v2.14.0
  over `backend/` and, through the `frontend-lint` stage, eslint; `test` runs
  the Go tests on the Debian Go image and, through the `frontend` stage, the
  frontend's unit tests and build. The builder stage waits on both, and without
  a `VERSION` build argument takes the version from `git describe` on the `.git`
  in the build context. Neither linter nor the tests run on the host for
  `make check`; `backend/script/lint` runs the root one, and
  `backend/.golangci.yml` is no longer checked against a sha256. prettier
  formats only the JavaScript, CSS, HTML and Markdown, so the shared
  `.golangci.yml` stays as fetched. `script/fmt` and `script/fmt-check` put
  `~/.local/bin` on `PATH`, which the shared workflow no longer does.
  `script/frontend-lint` and `script/frontend-check` are gone
 - 2026-10-07: the six Hetzner targets answer again, and failed checks are
  written to the browser console
  ([#114](https://git.eeqj.de/sneak/netwatch/issues/114)). The Hetzner
  speed-test servers close the connection without an answer when the URL has a
  query string, and every check added `?_cb=` and the time; checks now fetch
  each target's URL as written, with `cache: 'no-store'` as before. Each
  recorded check that fails writes one `console.error` line, and the same line
  to the debug log, with the target's name and URL, the time, what failed and
  how long the request took; a target that answers after failed checks writes
  one `console.info` line. Checks the page does not record write none, so the
  debug log no longer lists failures in the first round or in the recovery probe
 - 2026-10-04: the page's footer no longer says "IPv4 only"
  ([#111](https://git.eeqj.de/sneak/netwatch/issues/111)): each check is a
  `fetch`, the browser picks IPv4 or IPv6 for each WAN host, and the local
  targets are IPv4 addresses. The rest of the footer is unchanged
 - 2026-10-04: `README.md`, `TODO.md` and `test/viewport/README.md` say what the
  tree does (issue #24). The README's Getting Started leads with `make` targets;
  a new Backend section says what `netwatch-server` stores, its routes and how
  the image builds and runs it, and points to `backend/README.md` for its
  settings; the checks are GET requests; the 26 WAN hosts, the four health
  states, the summary's figures and the features the list lacked are described
  as the page has them; and its TODO section points here, as does the one in
  `backend/README.md`, whose open items moved to Future Steps. This file's
  Workflow branches from `next` and opens the PR against `next`, Status says
  where the repo stands, and Next Step and Future Steps hold only open work,
  linked to its issue where one exists. The viewport harness README names Node's
  test runner, not `vitest`
 - 2026-10-04: in `src/main.js` (issue #102), a target's min, max, median and
  average latency come from one list of its answers, through the same function
  the summary's figures use, so the median is written once. The latency color
  limits are one table in `CONFIG`, read by both the figure's and the
  sparkline's color. The health thresholds, the debug log's length, the gateway
  check's timeout, the recovery probe's number of hosts and interval, how often
  the rows are sorted and the delay before the first sparkline resize are
  `CONFIG` entries too. A unit test checks the summary's figures. Nothing the
  page does or shows changed; the footer's color legend still writes the limits
  out as text
 - 2026-10-04: password guesses at `/metrics` are rate limited (issue #104): each
  client address, resolved through `TRUSTED_PROXIES` as for reports, may make 60
  requests to `/metrics` a minute, counted by `go-chi/httprate` apart from its
  reports; past that it gets 429 and its basic auth credentials are not checked.
  The limit is a constant in `backend/internal/server/routes.go`. A test uses up
  one client's allowance on wrong passwords, gets 429 with the right one, and
  checks that another client behind the same nginx still gets in
 - 2026-10-04: the backend reports errors to Sentry (issue #95). With
  `SENTRY_DSN` set, it sets up `sentry-go` with the release `netwatch-server-`
  and its version, reports each panic in a handler through `sentryhttp`, the
  last of the middleware every request goes through, which panics again so the
  request still gets the 500 from the panic recovery, and waits up to 2 seconds
  on shutdown for Sentry to finish sending. A DSN Sentry refuses stops the start
  with an error naming `SENTRY_DSN`. With it empty, Sentry is not set up and
  nothing is sent to it
 - 2026-10-04: a target's name and URL and a debug log message show as the
  characters they are and are never read as HTML (issue #29): a host row escapes
  the name and URL it writes into its markup, and the debug log sets each line
@@ -514,18 +349,12 @@ Run `make frontend-viewport-test` in CI as its own step; it is not part of
 # Future Steps
- A backend test that posts a report to `POST /api/v1/reports` and checks the
+- Wire `script/frontend-viewport-test` into CI as its own step (deliberately not
-  compressed file it is written to
+  part of `make check` today; the decision has real CI-runtime cost and is
- A backend route that decompresses the stored reports and answers queries on
+  tracked separately)
-  them
+- Compliance top-up as one small commit: add .editorconfig and add the hooks
- Prometheus metrics for the backend's in-memory buffer: its size, the number of
+  target to the Makefile
-  flushes and the number of reports
+- After merge, confirm .gitea/workflows/check.yml is on main and CI is green
- A configurable host list (an environment variable or a config file)
+  (main always green policy)
- Export of the latency history (CSV or JSON)
+- Decide what to do with untracked resume.sh: commit it, gitignore it, or delete
- A notification when the health status changes to DEGRADED
+  it
 - `make test` within the 20-second target
  ([#118](https://git.eeqj.de/sneak/netwatch/issues/118)): every run compiles
  the Go standard library and every dependency with the race detector from
  nothing. Whether test builds may keep Go's compiled packages between runs
  waits on sneak's ruling in
  [prompts#113](https://git.eeqj.de/sneak/prompts/issues/113)
@@ -0,0 +1,2 @@
 node_modules
 .DS_Store
@@ -0,0 +1,6 @@
 /netwatch-server
 *.log
 *.out
 *.test
 .env
 data/
@@ -17,7 +17,6 @@ linters:
  disable:
    # Genuinely incompatible with project patterns
    - exhaustruct      # Requires all struct fields
    - exhaustruct_v5   # Requires all struct fields (successor to exhaustruct)
    - godot            # Requires comments to end with periods
    - wrapcheck        # Too verbose for internal packages
    - varnamelen       # Short names like db, id are idiomatic Go
@@ -0,0 +1,21 @@
 MIT License
 Copyright (c) 2026 @sneak (https://sneak.berlin)
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
 in the Software without restriction, including without limitation the rights
 to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 copies of the Software, and to permit persons to whom the Software is
 furnished to do so, subject to the following conditions:
 The above copyright notice and this permission notice shall be included in all
 copies or substantial portions of the Software.
 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 SOFTWARE.
@@ -0,0 +1,30 @@
 # Thin shims; the implementations live in backend/script/ (see the
 # Entrypoints section of README.md). There is no check, hooks or docker
 # target here: the root Makefile's check covers this directory, its
 # hooks target installs the repo's only pre-commit hook, and its docker
 # target builds the one image, which contains this backend.
 .PHONY: all build test lint fmt fmt-check run clean
 all: build
 build:
 	@script/build
 test:
 	@script/test
 lint:
 	@script/lint
 fmt:
 	@script/fmt
 fmt-check:
 	@script/fmt-check
 run:
 	@script/run
 clean:
 	@script/clean
@@ -0,0 +1,213 @@
 netwatch-server is an MIT-licensed Go HTTP backend by
 [@sneak](https://sneak.berlin) that receives telemetry reports from the NetWatch
 SPA and persists them as zstd-compressed JSONL files on disk.
 ## Getting Started
 From this directory:
 ```bash
 # Build and run locally
 make run
 ```
 From the repo root, whose `Dockerfile` builds the one image that ships this
 backend behind nginx (see [Container image](#container-image)):
 ```bash
 # Run tests, lint, and format check over the frontend and this backend
 make check
 # Build the image: nginx, the frontend and this backend
 make docker
 docker run -p 8080:8080 netwatch
 ```
 ## Entrypoints
 This directory follows the same
 [Scripts to Rule Them All](https://github.com/github/scripts-to-rule-them-all)
 pattern as the repo root: the targets in `backend/Makefile` are thin shims over
 `backend/script/`. The root `Dockerfile` runs them, and the root scripts call
 `test`, `fmt` and `fmt-check`:
 - `script/build` — compile the static `netwatch-server` binary with its version
  stamped in. The version is `VERSION` from the environment; when that is unset
  or empty, it falls back to `git describe` inside a git checkout, then to `dev`
 - `script/test` — run the Go tests with the race detector and coverage. Go's
  `-timeout 30s` bounds the tests, not their compile. If they fail, they run
  again with `-v` for the details, and the script fails. The race detector needs
  a C compiler
 - `script/lint` — check `.golangci.yml` against its pinned sha256, then run
  golangci-lint. It runs inside the golangci-lint image of the lint stage of the
  root `Dockerfile`; from a checkout, run `make lint` at the repo root, which
  builds that stage
 - `script/fmt` — format the Go sources (writes)
 - `script/fmt-check` — check Go formatting (read-only)
 - `script/run` — build and run the server locally
 - `script/clean` — remove build artifacts
 There is no `check`, `hooks` or `docker` target here: the root `make check`
 covers this directory, the root `make hooks` installs the repo's only pre-commit
 hook, and the root `make docker` builds the image that contains this backend.
 ## Rationale
 The NetWatch frontend collects latency measurements from the browser but has no
 way to persist or aggregate them. This backend provides a minimal
 `POST /api/v1/reports` endpoint that buffers incoming reports in memory and
 flushes them to compressed files on disk for later analysis.
 ## Design
 The server is structured as an `fx`-wired Go application under
 `cmd/netwatch-server/`. Internal packages in `internal/` follow standard Go
 project layout:
 - **`config`**: Loads configuration from environment variables and config files
  via Viper.
 - **`handlers`**: HTTP request handlers for the API (health check, report
  ingestion).
 - **`reportbuf`**: In-memory buffer that accumulates JSONL report lines and
  flushes to zstd-compressed files when the buffer reaches 10 MiB or every 60
  seconds.
 - **`server`**: Chi-based HTTP server with middleware wiring and route
  registration.
 - **`healthcheck`**, **`middleware`**, **`logger`**, **`globals`**: Supporting
  infrastructure.
 ### Configuration
 | Variable               | Default              | Description                                                                                              |
 | ---------------------- | -------------------- | -------------------------------------------------------------------------------------------------------- |
 | `BIND_ADDRESS`         | empty                | IP address to listen on; empty listens on every interface                                                |
 | `PORT`                 | `8080`               | HTTP listen port                                                                                         |
 | `DATA_DIR`             | `./data/reports`     | Directory for compressed reports                                                                         |
 | `DATA_DIR_MAX_BYTES`   | `1073741824` (1 GiB) | Most bytes of report files kept in `DATA_DIR`, oldest deleted first; see [Report limits](#report-limits) |
 | `DEBUG`                | `false`              | Enable debug logging                                                                                     |
 | `TRUSTED_PROXIES`      | loopback + RFC1918   | Comma-separated CIDRs whose `X-Forwarded-For` / `X-Real-IP` headers are trusted for client IP resolution |
 | `REPORTS_PER_MINUTE`   | `60`                 | Reports each client address may send a minute; see [Report limits](#report-limits)                       |
 | `CORS_ALLOWED_ORIGINS` | empty                | Comma-separated origins whose pages may call the API; see [CORS](#cors)                                  |
 | `METRICS_USERNAME`     | empty                | Basic auth user name for `/metrics`; see [Metrics](#metrics)                                             |
 | `METRICS_PASSWORD`     | empty                | Basic auth password for `/metrics`; see [Metrics](#metrics)                                              |
 `TRUSTED_PROXIES` defaults to
 `127.0.0.1/32,::1/128,10.0.0.0/8,172.16.0.0/12,192.168.0.0/16`. The loopback
 entries cover a reverse proxy on the same host. A request whose direct peer is
 outside this set has its forwarded headers ignored, and the direct peer is
 logged and rate-limited instead. The container image does not use this default;
 see [Container image](#container-image).
 A variable set to a value the server cannot use, such as `PORT=abc`,
 `DEBUG=maybe` or a `BIND_ADDRESS` that is not an IP address, stops it from
 starting, with an error naming the variable. An empty variable counts as unset.
 ### Container image
 The root `Dockerfile` builds one image in which nginx listens on the public port
 8080, serves the frontend, and proxies `/api/`, `/.well-known/healthcheck` and
 `/metrics` to this server. The image's entrypoint, `bin/entrypoint.sh`, starts
 the server as user `netwatch` (uid 1000) with `BIND_ADDRESS=127.0.0.1` and
 `PORT=8081`, so only nginx reaches it, and with `TRUSTED_PROXIES=127.0.0.1/32`,
 so it takes the client address nginx passes on and no other. `DATA_DIR` is
 `/data/reports`, on the `/data` volume; before starting the server, the
 entrypoint creates it and gives it and `/data` to `netwatch` with
 `netwatch-server prepare-data-dir`, which acts on nothing outside `/data`. nginx
 replaces the security headers this server sets with those in the root
 `security-headers.conf`, so those are what clients of the image see.
 The container's own `TRUSTED_PROXIES` goes to nginx instead: IP addresses or
 CIDRs, separated by commas, of the reverse proxies in front of the container.
 nginx takes the client address from `X-Forwarded-For` only on a request from one
 of them. Unset or empty, nginx trusts no proxy, and the client address is the
 one each request comes from, so every client behind a proxy shares one rate
 limit. An entry that is not an IP address or CIDR, such as a hostname or
 `1.2.3`, stops the container at start with an error naming `TRUSTED_PROXIES`:
 the entrypoint checks each entry with `netwatch-server check-cidr`, which parses
 it as this server parses its own `TRUSTED_PROXIES`.
 ### Report storage
 Reports are written as `reports-<timestamp>-<number>.jsonl.zst` files in
 `DATA_DIR`. The timestamp is in UTC to the millisecond, so the names sort by
 time. The number starts at 1 when the server starts and goes up by one for each
 file the server starts to write, so two files written in the same millisecond
 still get different names. A failed write uses up its number and leaves a gap in
 the numbers: its file, if it was created, is removed. The file stays, counted
 toward `DATA_DIR_MAX_BYTES` from the next start, only if removing it fails too.
 Each file contains one JSON object per line, compressed with zstd. Files are
 created with `O_EXCL` to prevent overwrites.
 ### Report limits
 `POST /api/v1/reports` takes reports from anyone who can reach it, without
 credentials, so it is bounded instead. Both refusals below answer with the same
 `{"status":"error"}` body as any other error.
 - **Rate limit.** Each client address, resolved through `TRUSTED_PROXIES`, may
  send `REPORTS_PER_MINUTE` reports a minute; past that it gets 429 with
  `Retry-After: 60`. The minute slides: reports from the minute before still
  count, fading out over the current one, so an address is sure never to be
  refused only while it sends at most half of `REPORTS_PER_MINUTE` in any 60
  seconds. The page sends one report a minute from each open tab, so the default
  of 60 refuses nothing from up to 30 tabs behind one address, such as a
  household or an office sharing it, however their reports bunch up. Report
  responses also carry `X-RateLimit-Limit`, `X-RateLimit-Remaining` and
  `X-RateLimit-Reset` headers.
 - **Size cap.** The report files in `DATA_DIR` may total at most
  `DATA_DIR_MAX_BYTES`, counting the files already there at start. Reports
  waiting in memory count at their uncompressed size until they are written;
  those lost to a failed write stop counting, and the part of its file written
  is removed. When a report would take the total past the cap, the oldest report
  files are deleted to make room, and each deletion is logged with the file's
  name and size; a file still being written is never deleted. A report is
  refused with 507, and nothing of it is stored, only when the reports waiting
  to be written fill the cap on their own, and then no file is deleted. At
  start, report files past the cap, as after lowering it, are deleted the same
  way. So the cap is how much of the newest reports is kept: the default of 1
  GiB is small enough for any host; set it to the space you can give `DATA_DIR`.
 ### CORS
 The page calls the API from the origin it is served from, so by default the
 server sends no CORS headers, and browsers let no other origin's pages call it.
 To serve the page from elsewhere, list that origin in `CORS_ALLOWED_ORIGINS`
 (for example `https://netwatch.example.com`); pages from a listed origin may
 `GET` and `POST` with a `Content-Type` header. Each entry must be a plain
 origin, `scheme://host` with an optional `:port`, as browsers send it: no path,
 not even a trailing `/`, and no `*`. Any other entry stops the server from
 starting, with an error naming `CORS_ALLOWED_ORIGINS`.
 ### Metrics
 With both `METRICS_USERNAME` and `METRICS_PASSWORD` set, the server serves
 Prometheus metrics at `GET /metrics` to requests with those as their basic auth
 credentials, and answers any other with 401. For each request that reaches the
 health check or `POST /api/v1/reports`, those the rate limit refuses included,
 the metrics record its duration and response size, labelled with its path,
 method and status; they also count those requests in progress, and include Go's
 runtime and process metrics. No other request is recorded: not those to
 `/metrics` itself, and not those answered before they reach either route, such
 as a CORS preflight, or a request refused with 404 for a path no route has, 405
 for a method its route does not take, or 413 for declaring a body length over
 the 1 MiB limit. A report whose body goes over the limit without declaring its
 length reaches the route, is answered 413 there, and is recorded with that
 status. Clients can make up any number of paths and methods, and each would add
 labels to the metrics for as long as the server runs. With neither set, nothing
 is recorded and `/metrics` answers 404. One without the other stops the server
 from starting, with an error naming both; so does a `METRICS_USERNAME`
 containing `:`, which basic auth cannot carry, with an error naming it.
 ## TODO
 - Add integration test that POSTs a report and verifies the compressed output
 - Add report decompression/query endpoint
 - Add metrics (Prometheus) for buffer size, flush count, report count
 ## License
 MIT. See [LICENSE](LICENSE).
 ## Author
 [@sneak](https://sneak.berlin)
@@ -115,28 +115,6 @@ func TestMalformedConfigFileStopsTheStart(t *testing.T) {
 	}
 }
 // TestRefusedSentryDSNStopsTheStart: a SENTRY_DSN that Sentry refuses
 // stops the start, and the error, naming SENTRY_DSN, is logged as JSON.
 func TestRefusedSentryDSNStopsTheStart(t *testing.T) {
 	t.Setenv("SENTRY_DSN", "not-a-dsn")
 	ctx, cancel := context.WithTimeout(t.Context(), childTimeout)
 	defer cancel()
 	child, stdout, stderr := startServer(ctx, t, t.TempDir(), freePort(ctx, t))
 	err := child.Wait()
 	if child.ProcessState.ExitCode() != 1 {
 		t.Fatalf("server exit = %v, want exit status 1", err)
 	}
 	requireJSONLines(t, stdout, stderr)
 	if !strings.Contains(stdout.String(), "SENTRY_DSN") {
 		t.Fatalf("no error naming SENTRY_DSN in stdout:\n%s", stdout)
 	}
 }
 // startServer runs main() in a child process listening on
 // 127.0.0.1:port, with home as its HOME and working directory and its
 // data directory in home, so it touches nothing outside home. Its
@@ -2,12 +2,8 @@ module sneak.berlin/go/netwatch
 go 1.25.5
 // npm packages can ship .go files, which ./... and all would include.
 ignore ./node_modules
 require (
 	github.com/99designs/basicauth-go v0.0.0-20230316000542-bf6f9cbbf0f8
 	github.com/getsentry/sentry-go v0.49.0
 	github.com/go-chi/chi/v5 v5.2.5
 	github.com/go-chi/cors v1.2.2
 	github.com/go-chi/httprate v0.16.0
@@ -4,22 +4,18 @@ github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=
 github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6rlkpw=
 github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
 github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
-github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc h1:U9qPSI2PIWSS1VwoXQT9A3Wy9MM3WgvqSxFWenqJduM=
+github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
-github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/frankban/quicktest v1.14.6 h1:7Xjx+VpznH+oBnejlPUj8oUpdxnVs4f8XU8WnHkI4W8=
 github.com/frankban/quicktest v1.14.6/go.mod h1:4ptaffx2x8+WTWXmUCuVU6aPUX1/Mz7zb5vbUoiM6w0=
 github.com/fsnotify/fsnotify v1.9.0 h1:2Ml+OJNzbYCTzsxtv8vKSFD9PbJjmhYF14k/jKC7S9k=
 github.com/fsnotify/fsnotify v1.9.0/go.mod h1:8jBTzvmWwFyi3Pb8djgCCO5IBqzKJ/Jwo8TRcHyHii0=
 github.com/getsentry/sentry-go v0.49.0 h1:Ehejknu1l023Ub7QoRBVLAI7g3Jnhqku4oWx4B4Sh5s=
 github.com/getsentry/sentry-go v0.49.0/go.mod h1:nuMJAoCfe1u0Bts2ocyNI+TW8HT84vRMqwA5Qq/SKUI=
 github.com/go-chi/chi/v5 v5.2.5 h1:Eg4myHZBjyvJmAFjFvWgrqDTXFyOzjj7YIm3L3mu6Ug=
 github.com/go-chi/chi/v5 v5.2.5/go.mod h1:X7Gx4mteadT3eDOMTsXzmI4/rwUpOwBHLpAfupzFJP0=
 github.com/go-chi/cors v1.2.2 h1:Jmey33TE+b+rB7fT8MUy1u0I4L+NARQlK6LhzKPSyQE=
 github.com/go-chi/cors v1.2.2/go.mod h1:sSbTewc+6wYHBBCW7ytsFSn836hqM7JxpglAy2Vzc58=
 github.com/go-chi/httprate v0.16.0 h1:8V5DH9j6pSK6UQoBsTpvMyFxycqaKEIToyPKzHJjUa8=
 github.com/go-chi/httprate v0.16.0/go.mod h1:A8lo+qRhk+s9LiuP5saS7XCGDXRXMcrueq0NfIuCa/I=
 github.com/go-errors/errors v1.4.2 h1:J6MZopCL4uSllY1OfXM374weqZFFItUbrImctkmUxIA=
 github.com/go-errors/errors v1.4.2/go.mod h1:sIVyrIiJhuEF+Pj9Ebtd6P/rEYROXFi3BopGUQ5a5Og=
 github.com/go-viper/mapstructure/v2 v2.4.0 h1:EBsztssimR/CONLSZZ04E8qAkxNYq4Qp9LvH92wZUgs=
 github.com/go-viper/mapstructure/v2 v2.4.0/go.mod h1:oJDH3BJKyqBA2TXFhDsKDGDTlndYOZ6rGS0BRZIxGhM=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
@@ -40,12 +36,8 @@ github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
 github.com/pelletier/go-toml/v2 v2.2.4 h1:mye9XuhQ6gvn5h28+VilKrrPoQVanw5PMw/TB0t5Ec4=
 github.com/pelletier/go-toml/v2 v2.2.4/go.mod h1:2gIqNv+qfxSVS7cM2xJQKtLSTLUE9V8t9Stt+h56mCY=
-github.com/pingcap/errors v0.11.4 h1:lFuQV/oaUMGcD2tqt+01ROSmJs75VG1ToEOkZIZ4nE4=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
-github.com/pingcap/errors v0.11.4/go.mod h1:Oi8TUi2kEtXXLMJk9l1cGmz20kV3TaQ0usTwv5KuLY8=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/pkg/errors v0.9.1 h1:FEBLx1zS214owpjy7qsBeixbURkuhQAwrK5UwLGTwt4=
 github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/prometheus/client_golang v1.24.1 h1:JnJkREXzWxUdCuPFpIWZiPispT9xVV59uiuyR2bPlnU=
 github.com/prometheus/client_golang v1.24.1/go.mod h1:F+oSRECHg4sse5ucfYpYDeIv/hu68Zo0uoHKetWnzcE=
 github.com/prometheus/client_model v0.6.2 h1:oBsgwpGs7iVziMvrGhE53c/GrLUsZdHnqNwqPLxwZyk=
@@ -54,8 +46,8 @@ github.com/prometheus/common v0.70.1 h1:1HvjP4D5oL3t8RsPlwxA9onvvStjtIHYE5XuuwOi
 github.com/prometheus/common v0.70.1/go.mod h1:VdFUQDMZK3VLkurFUVhia6uys/0suUp86TJz5qbJRhc=
 github.com/prometheus/procfs v0.21.1 h1:GljZCt+zSTS+NZq88cyQ1LjZ+RCHp3uVuabBWA5+OJI=
 github.com/prometheus/procfs v0.21.1/go.mod h1:aB55Cww9pdSJVHk0hUf0inxWyyjPogFIjmHKYgMKmtY=
-github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
+github.com/rogpeppe/go-internal v1.9.0 h1:73kH8U+JUqXU8lRuOHeVHaa/SZPifC7BkcraZVejAe8=
-github.com/rogpeppe/go-internal v1.14.1/go.mod h1:MaRKkUm5W0goXpeCfT7UZI6fk/L7L7so1lCWt35ZSgc=
+github.com/rogpeppe/go-internal v1.9.0/go.mod h1:WtVeX8xhTBvf0smdhujwtBcq4Qrzq/fJaraNFVN+nFs=
 github.com/sagikazarmark/locafero v0.11.0 h1:1iurJgmM9G3PA/I+wWYIOw/5SyBtxapeHDcg+AAIFXc=
 github.com/sagikazarmark/locafero v0.11.0/go.mod h1:nVIGvgyzw595SUSUE6tvCp3YYTeHs15MvlmU87WwIik=
 github.com/slok/go-http-metrics v0.13.0 h1:lQDyJJx9wKhmbliyUsZ2l6peGnXRHjsjoqPt5VYzcP8=
@@ -101,7 +93,7 @@ golang.org/x/text v0.40.0/go.mod h1:hpnzDAfGV753zIKo+wk3u1bVKCGPbrnF7+7LBF/UHVY=
 google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
 google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
+gopkg.in/check.v1 v1.0.0-20190902080502-41f04d3bba15 h1:YR8cESwS4TdDjEe65xsg0ogRM/Nc3DYOhEAlW+xobZo=
-gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
+gopkg.in/check.v1 v1.0.0-20190902080502-41f04d3bba15/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
@@ -9,7 +9,6 @@ import (
 	"math"
 	"net/netip"
 	"net/url"
 	"os"
 	"strconv"
 	"strings"
@@ -22,14 +21,16 @@ import (
 )
 // defaultTrustedProxies lists the networks whose forwarded
-// headers are honoured when TRUSTED_PROXIES is unset: the
+// headers are honoured by default: IPv4 and IPv6 loopback,
-// RFC 1918 ranges. The container image does not use it:
+// for a reverse proxy on the same host, and the RFC1918
 // ranges. The container image does not use it:
 // bin/entrypoint.sh gives the server 127.0.0.1/32, since
 // nginx is its only client there.
-const defaultTrustedProxies = "10.0.0.0/8,172.16.0.0/12,192.168.0.0/16"
+const defaultTrustedProxies = "127.0.0.1/32,::1/128," +
 	"10.0.0.0/8,172.16.0.0/12,192.168.0.0/16"
-// Default limits on stored reports; README.md gives the reasons for
+// Default limits on stored reports; backend/README.md gives the
-// these values.
+// reasons for these values.
 const (
 	defaultReportsPerMinute = 60
 	defaultDataDirMaxBytes  = 1 << 30 // 1 GiB
@@ -134,14 +135,6 @@ func New(
 			viper.GetString("DEBUG"), errNotBool)
 	}
 	// An empty TRUSTED_PROXIES trusts no proxy, but viper reads an
 	// empty environment variable as unset and would give the default,
 	// so the environment is read first.
 	trustedProxies, set := os.LookupEnv("TRUSTED_PROXIES")
 	if !set {
 		trustedProxies = viper.GetString("TRUSTED_PROXIES")
 	}
 	s := &Config{
 		BindAddress:        viper.GetString("BIND_ADDRESS"),
 		CORSAllowedOrigins: splitList(viper.GetString("CORS_ALLOWED_ORIGINS")),
@@ -153,7 +146,7 @@ func New(
 		Port:               port,
 		ReportsPerMinute:   viper.GetInt("REPORTS_PER_MINUTE"),
 		SentryDSN:          viper.GetString("SENTRY_DSN"),
-		TrustedProxies:     splitList(trustedProxies),
+		TrustedProxies:     splitList(viper.GetString("TRUSTED_PROXIES")),
 		log:                log,
 		params:             &params,
 	}
@@ -1,8 +1,6 @@
 package config_test
 import (
 	"os"
 	"slices"
 	"strings"
 	"testing"
@@ -31,62 +29,6 @@ func requireConfigError(t *testing.T, setting string) {
 	}
 }
 // requireTrustedProxies builds the config as main does and fails the
 // test unless its TrustedProxies are want.
 func requireTrustedProxies(t *testing.T, want []string) {
 	t.Helper()
 	var cfg *config.Config
 	app := fx.New(
 		fx.NopLogger,
 		fx.Provide(globals.New, logger.New, config.New),
 		fx.Populate(&cfg),
 	)
 	err := app.Err()
 	if err != nil {
 		t.Fatalf("config error = %v", err)
 	}
 	if !slices.Equal(cfg.TrustedProxies, want) {
 		t.Fatalf("TrustedProxies = %q, want %q", cfg.TrustedProxies, want)
 	}
 }
 // TestTrustedProxiesUnsetTrustsRFC1918: unset, the server trusts the
 // RFC 1918 ranges and nothing else, loopback included.
 func TestTrustedProxiesUnsetTrustsRFC1918(t *testing.T) {
 	// t.Setenv first, so the variable is put back as it was after the
 	// test.
 	t.Setenv("TRUSTED_PROXIES", "")
 	err := os.Unsetenv("TRUSTED_PROXIES")
 	if err != nil {
 		t.Fatal(err)
 	}
 	requireTrustedProxies(t,
 		[]string{"10.0.0.0/8", "172.16.0.0/12", "192.168.0.0/16"})
 }
 // TestTrustedProxiesEmptyTrustsNothing: set and empty, the server
 // trusts no proxy. viper alone would read the empty variable as unset
 // and give the default.
 func TestTrustedProxiesEmptyTrustsNothing(t *testing.T) {
 	t.Setenv("TRUSTED_PROXIES", "")
 	requireTrustedProxies(t, []string{})
 }
 // TestTrustedProxiesListReplacesDefault: a list is used as given, with
 // none of the default ranges added.
 func TestTrustedProxiesListReplacesDefault(t *testing.T) {
 	t.Setenv("TRUSTED_PROXIES", "172.17.0.1/32, 2001:db8::/32")
 	requireTrustedProxies(t, []string{"172.17.0.1/32", "2001:db8::/32"})
 }
 // TestSettingsLoadAsGiven: valid values pass the checks and are used
 // as given. bin/entrypoint.sh starts the server with these
 // BIND_ADDRESS and PORT values.
@@ -5,9 +5,7 @@ package handlers
 import (
 	"context"
 	"encoding/json"
 	"errors"
 	"log/slog"
 	"mime"
 	"net/http"
 	"sneak.berlin/go/netwatch/internal/globals"
@@ -20,10 +18,6 @@ import (
 const jsonContentType = "application/json; charset=utf-8"
 // errNotJSON is what decodeJSON returns for a request whose
 // Content-Type is not application/json.
 var errNotJSON = errors.New("content type is not application/json")
 // reportAppender is the subset of the report buffer the handlers
 // depend on. Defining it here keeps the storage failure path
 // exercisable with a stub in tests.
@@ -86,25 +80,14 @@ func (s *Handlers) respondJSON(
 	}
 }
-// decodeJSON decodes the request body into v. It returns errNotJSON,
+// decodeJSON decodes the request body into v. The body is
-// reading nothing, unless the request's Content-Type is
+// expected to already be bounded by the body-size middleware, so
-// application/json, with or without parameters such as charset. The
+// a caller can distinguish an over-limit body from malformed
-// body is expected to already be bounded by the body-size middleware,
+// JSON by testing the returned error for *http.MaxBytesError.
 // so a caller can distinguish an over-limit body from malformed JSON
 // by testing the returned error for *http.MaxBytesError.
 func (s *Handlers) decodeJSON(
 	_ http.ResponseWriter,
 	r *http.Request,
 	v any,
 ) error {
 	// A page on another origin can POST text/plain or a form without
 	// a CORS preflight, so its visitors' browsers would send it
 	// whatever the CORS settings. A JSON POST waits for the
 	// preflight, so only JSON is taken.
 	mediaType, _, err := mime.ParseMediaType(r.Header.Get("Content-Type"))
 	if err != nil || mediaType != "application/json" {
 		return errNotJSON
 	}
 	return json.NewDecoder(r.Body).Decode(v)
 }
@@ -30,7 +30,7 @@ type report struct {
 }
 // HandleReport returns a handler that accepts telemetry
-// reports from NetWatch clients, sent as application/json.
+// reports from NetWatch clients.
 func (s *Handlers) HandleReport() http.HandlerFunc {
 	type response struct {
 		Status string `json:"status"`
@@ -66,15 +66,9 @@ func (s *Handlers) HandleReport() http.HandlerFunc {
 }
 // decodeErrorStatus logs a report decode failure and returns the
-// status to send: 415 when the report was not sent as JSON, 413 when
+// status to send: 413 when the body exceeded the size limit,
-// the body exceeded the size limit, otherwise 400 for malformed JSON.
+// otherwise 400 for malformed JSON.
 func (s *Handlers) decodeErrorStatus(err error) int {
 	if errors.Is(err, errNotJSON) {
 		s.log.Warn("report refused: Content-Type is not application/json")
 		return http.StatusUnsupportedMediaType
 	}
 	var tooLarge *http.MaxBytesError
 	if errors.As(err, &tooLarge) {
 		s.log.Warn("report body too large", "limit_bytes", tooLarge.Limit)
@@ -93,7 +87,7 @@ func (s *Handlers) decodeErrorStatus(err error) int {
 // appendErrorStatus logs a failure to store a report and returns
 // the status to send: 507 when the reports waiting to be written fill
-// the size cap, otherwise 500, as when report files cannot be written.
+// the size cap, otherwise 500.
 func (s *Handlers) appendErrorStatus(err error) int {
 	if errors.Is(err, reportbuf.ErrFull) {
 		s.log.Warn("report refused: " +
@@ -27,17 +27,6 @@ type stubAppender struct {
 func (s stubAppender) Append(any) error { return s.err }
 // recordingAppender keeps every report the handler stores.
 type recordingAppender struct {
 	reports []any
 }
 func (a *recordingAppender) Append(v any) error {
 	a.reports = append(a.reports, v)
 	return nil
 }
 func newTestHandlers(buf stubAppender, out io.Writer) *handlers.Handlers {
 	return handlers.NewForTest(buf, slog.New(slog.NewJSONHandler(out, nil)))
 }
@@ -94,7 +83,6 @@ func TestHandleReportAcceptsValidReports(t *testing.T) {
 				http.MethodPost, "/api/v1/reports",
 				strings.NewReader(tt.body),
 			)
 			req.Header.Set("Content-Type", "application/json")
 			h.HandleReport().ServeHTTP(rec, req)
@@ -115,89 +103,6 @@ func TestHandleReportAcceptsValidReports(t *testing.T) {
 	}
 }
 // TestHandleReportStoresOnlyJSON checks that a report is stored only
 // when sent as application/json, as the page sends it. One sent as
 // text/plain or as a form, which a page on another origin can send
 // without a CORS preflight, is refused with 415.
 func TestHandleReportStoresOnlyJSON(t *testing.T) {
 	t.Parallel()
 	tests := []struct {
 		name        string
 		contentType string
 		wantStatus  int
 		wantStored  int
 	}{
 		{
 			name:        "text/plain",
 			contentType: "text/plain;charset=UTF-8",
 			wantStatus:  http.StatusUnsupportedMediaType,
 			wantStored:  0,
 		},
 		{
 			name:        "a urlencoded form",
 			contentType: "application/x-www-form-urlencoded",
 			wantStatus:  http.StatusUnsupportedMediaType,
 			wantStored:  0,
 		},
 		{
 			name:        "a multipart form",
 			contentType: "multipart/form-data; boundary=x",
 			wantStatus:  http.StatusUnsupportedMediaType,
 			wantStored:  0,
 		},
 		{
 			name:        "no Content-Type",
 			contentType: "",
 			wantStatus:  http.StatusUnsupportedMediaType,
 			wantStored:  0,
 		},
 		{
 			name:        "application/json",
 			contentType: "application/json",
 			wantStatus:  http.StatusOK,
 			wantStored:  1,
 		},
 		{
 			name:        "application/json with a charset",
 			contentType: "application/json;charset=UTF-8",
 			wantStatus:  http.StatusOK,
 			wantStored:  1,
 		},
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			t.Parallel()
 			var stored recordingAppender
 			h := handlers.NewForTest(&stored, slog.New(slog.DiscardHandler))
 			rec := httptest.NewRecorder()
 			req := httptest.NewRequestWithContext(t.Context(),
 				http.MethodPost, "/api/v1/reports",
 				strings.NewReader(`{"clientId":"c1","hosts":[]}`),
 			)
 			if tt.contentType != "" {
 				req.Header.Set("Content-Type", tt.contentType)
 			}
 			h.HandleReport().ServeHTTP(rec, req)
 			if rec.Code != tt.wantStatus {
 				t.Errorf("status = %d, want %d", rec.Code, tt.wantStatus)
 			}
 			if len(stored.reports) != tt.wantStored {
 				t.Errorf("reports stored = %d, want %d",
 					len(stored.reports), tt.wantStored)
 			}
 		})
 	}
 }
 func TestHandleReportStorageFailureIsNon2xx(t *testing.T) {
 	t.Parallel()
@@ -208,7 +113,6 @@ func TestHandleReportStorageFailureIsNon2xx(t *testing.T) {
 		http.MethodPost, "/api/v1/reports",
 		strings.NewReader(`{"clientId":"c1","hosts":[]}`),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
@@ -234,7 +138,6 @@ func TestHandleReportFullIs507(t *testing.T) {
 		http.MethodPost, "/api/v1/reports",
 		strings.NewReader(`{"clientId":"c1","hosts":[]}`),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
@@ -258,7 +161,6 @@ func TestHandleReportMalformedJSONIs400(t *testing.T) {
 		http.MethodPost, "/api/v1/reports",
 		strings.NewReader(`{not json`),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
@@ -283,8 +185,6 @@ func TestHandleReportOversizeIs413(t *testing.T) {
 		http.MethodPost, "/api/v1/reports",
 		strings.NewReader(`{"clientId":"`+strings.Repeat("x", 200)+`"}`),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	// No declared length, so only the middleware's read cap can
 	// stop this body.
 	req.ContentLength = -1
@@ -313,7 +213,6 @@ func TestHandleReportDoesNotLogRawGeo(t *testing.T) {
 			`{"clientId":"c1","geo":{"raw":"`+sentinel+`"},"hosts":[]}`,
 		),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
@@ -346,7 +245,6 @@ func TestHandleReportLogsClientIDCutToBound(t *testing.T) {
 			`{"clientId":"`+long+`","timestamp":"`+long+`","hosts":[]}`,
 		),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
@@ -386,7 +284,6 @@ func TestHandleReportDecodeErrorLogIsBounded(t *testing.T) {
 		http.MethodPost, "/api/v1/reports",
 		strings.NewReader(`{"hosts":[{"history":[{"t":`+huge+`}]}]}`),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
@@ -297,16 +297,12 @@ func (s *Middleware) MaxBodyBytes(
 // Recoverer returns middleware that recovers from a panic in a
 // downstream handler, logs the panic and stack trace through
-// slog, and responds 500 with no body, unless a status was already
+// slog, and responds 500 with no body. http.ErrAbortHandler is
-// written, as Write500OnPanic writes one for a handler's panic.
+// re-panicked so the server can abort the response as intended.
 // http.ErrAbortHandler is re-panicked so the server can abort the
 // response as intended.
 func (s *Middleware) Recoverer() func(http.Handler) http.Handler {
 	return func(next http.Handler) http.Handler {
 		return http.HandlerFunc(
 			func(w http.ResponseWriter, r *http.Request) {
 				ww := middleware.NewWrapResponseWriter(w, r.ProtoMajor)
 				defer func() {
 					rec := recover()
 					if rec == nil {
@@ -324,43 +320,10 @@ func (s *Middleware) Recoverer() func(http.Handler) http.Handler {
 						"stack", string(debug.Stack()),
 					)
-					if ww.Status() == 0 {
+					w.WriteHeader(http.StatusInternalServerError)
 						ww.WriteHeader(http.StatusInternalServerError)
 					}
 				}()
-				next.ServeHTTP(ww, r)
+				next.ServeHTTP(w, r)
 			},
 		)
 	}
 }
 // Write500OnPanic returns middleware that writes 500 when a downstream
 // handler panics before writing a status, and lets the panic go on to
 // Sentry and Recoverer. Recoverer, which comes first, would write its
 // 500 only after the request log and the metrics had recorded the
 // status they saw; placed after each of them, this makes them record
 // the 500.
 func (s *Middleware) Write500OnPanic() func(http.Handler) http.Handler {
 	return func(next http.Handler) http.Handler {
 		return http.HandlerFunc(
 			func(w http.ResponseWriter, r *http.Request) {
 				ww := middleware.NewWrapResponseWriter(w, r.ProtoMajor)
 				defer func() {
 					rec := recover()
 					if rec == nil {
 						return
 					}
 					if ww.Status() == 0 {
 						ww.WriteHeader(http.StatusInternalServerError)
 					}
 					panic(rec)
 				}()
 				next.ServeHTTP(ww, r)
 			},
 		)
 	}
@@ -4,7 +4,6 @@ import (
 	"bytes"
 	"encoding/json"
 	"errors"
 	"log"
 	"log/slog"
 	"net/http"
 	"net/http/httptest"
@@ -17,9 +16,7 @@ import (
 	"sneak.berlin/go/netwatch/internal/logger"
 	"sneak.berlin/go/netwatch/internal/middleware"
 	"github.com/go-chi/chi/v5"
 	chimiddleware "github.com/go-chi/chi/v5/middleware"
 	"github.com/prometheus/client_golang/prometheus"
 )
 const (
@@ -326,172 +323,6 @@ func TestRecovererRepanicsOnAbortHandler(t *testing.T) {
 	}
 }
 // TestHandlerPanicIsLoggedAndCountedAs500 sends a request to a handler
 // that panics through the middleware SetupRoutes puts around a route
 // with metrics, in its order, leaving out those that write no status
 // for a panic and Sentry, which TestSentry covers. The request log line
 // and the metrics must give the 500 the client gets, and the panic must
 // still reach Recoverer, which logs it.
 func TestHandlerPanicIsLoggedAndCountedAs500(t *testing.T) {
 	t.Parallel()
 	var logbuf bytes.Buffer
 	mw := middleware.NewWithLogger(
 		slog.New(slog.NewJSONHandler(&logbuf, nil)),
 	)
 	registry := prometheus.NewRegistry()
 	handler := chi.Chain(
 		mw.Recoverer(),
 		chimiddleware.RequestID,
 		mw.Logging(),
 		mw.Write500OnPanic(),
 		mw.Metrics(registry),
 		mw.Write500OnPanic(),
 	).HandlerFunc(func(http.ResponseWriter, *http.Request) {
 		panic("boom")
 	})
 	rec := httptest.NewRecorder()
 	req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, "/", http.NoBody)
 	handler.ServeHTTP(rec, req)
 	if rec.Code != http.StatusInternalServerError {
 		t.Fatalf("status = %d, want %d",
 			rec.Code, http.StatusInternalServerError)
 	}
 	request := findLogLine(t, logbuf.Bytes(), "request")
 	if request.Status != http.StatusInternalServerError {
 		t.Errorf("request log status = %d, want %d",
 			request.Status, http.StatusInternalServerError)
 	}
 	recovered := findLogLine(t, logbuf.Bytes(), "panic recovered")
 	if recovered.Panic != "boom" {
 		t.Errorf("panic recovered line has panic %q, want %q",
 			recovered.Panic, "boom")
 	}
 	for _, name := range []string{
 		"http_request_duration_seconds", "http_response_size_bytes",
 	} {
 		codes := metricCodes(t, registry, name)
 		if len(codes) != 1 || codes[0] != "500" {
 			t.Errorf("%s code labels = %v, want [500]", name, codes)
 		}
 	}
 }
 // loggedLine holds the fields the tests read from a JSON log line.
 type loggedLine struct {
 	Msg    string `json:"msg"`
 	Status int    `json:"status"`
 	Panic  string `json:"panic"`
 }
 // findLogLine returns the first of the JSON log lines in logged whose
 // message is msg.
 func findLogLine(t *testing.T, logged []byte, msg string) loggedLine {
 	t.Helper()
 	dec := json.NewDecoder(bytes.NewReader(logged))
 	for {
 		var line loggedLine
 		err := dec.Decode(&line)
 		if err != nil {
 			t.Fatalf("no %q line in the log: %v (%q)", msg, err, logged)
 		}
 		if line.Msg == msg {
 			return line
 		}
 	}
 }
 // metricCodes returns the code label of each series of the metric
 // family name in registry.
 func metricCodes(
 	t *testing.T,
 	registry *prometheus.Registry,
 	name string,
 ) []string {
 	t.Helper()
 	families, err := registry.Gather()
 	if err != nil {
 		t.Fatalf("gather metrics: %v", err)
 	}
 	var codes []string
 	for _, family := range families {
 		if family.GetName() != name {
 			continue
 		}
 		for _, metric := range family.GetMetric() {
 			for _, label := range metric.GetLabel() {
 				if label.GetName() == "code" {
 					codes = append(codes, label.GetValue())
 				}
 			}
 		}
 	}
 	return codes
 }
 // TestRecovererWritesNoSecondStatus checks, on a real server, that
 // Recoverer does not write its 500 again after Write500OnPanic has
 // written one, which the server would log as a superfluous
 // WriteHeader call.
 func TestRecovererWritesNoSecondStatus(t *testing.T) {
 	t.Parallel()
 	var serverLog bytes.Buffer
 	mw := middleware.NewWithLogger(slog.New(slog.DiscardHandler))
 	server := httptest.NewUnstartedServer(chi.Chain(
 		mw.Recoverer(),
 		mw.Write500OnPanic(),
 	).HandlerFunc(func(http.ResponseWriter, *http.Request) {
 		panic("boom")
 	}))
 	server.Config.ErrorLog = log.New(&serverLog, "", 0)
 	server.Start()
 	req, err := http.NewRequestWithContext(t.Context(),
 		http.MethodGet, server.URL, http.NoBody)
 	if err != nil {
 		t.Fatalf("new request: %v", err)
 	}
 	resp, err := server.Client().Do(req)
 	if err != nil {
 		t.Fatalf("request: %v", err)
 	}
 	_ = resp.Body.Close()
 	// Close waits for the handler to return, so whatever the server
 	// logs for this request is in serverLog after it.
 	server.Close()
 	if resp.StatusCode != http.StatusInternalServerError {
 		t.Errorf("status = %d, want %d",
 			resp.StatusCode, http.StatusInternalServerError)
 	}
 	if serverLog.Len() != 0 {
 		t.Errorf("server logged %q", serverLog.String())
 	}
 }
 // TestLoggingCutsRequestStringsToBound sends an over-long URL and
 // over-long header values, and checks the request log writes each
 // one cut to logger.MaxLoggedFieldBytes.
@@ -512,7 +343,7 @@ func TestLoggingCutsRequestStringsToBound(t *testing.T) {
 		http.MethodGet, "/"+long, http.NoBody)
 	req.Header.Set("User-Agent", long)
 	req.Header.Set("Referer", long)
-	req.Header.Set("X-Request-ID", long)
+	req.Header.Set("X-Request-Id", long)
 	handler.ServeHTTP(httptest.NewRecorder(), req)
@@ -91,15 +91,6 @@ type Buffer struct {
 	// of the report files in dataDir, plus the reports waiting to
 	// be written to one at their uncompressed size.
 	usedBytes int64
 	// writeErr is the error of the last write of a report file to
 	// finish, nil if it succeeded or none has. While it is set, Append
 	// refuses reports, and the reports of that write wait in buf for
 	// the next flush.
 	writeErr error
 	// writers are the goroutines that write report files: flushLoop,
 	// and each one Append starts at flushSizeThreshold. OnStop waits
 	// for them before the final flush.
 	writers sync.WaitGroup
 }
 // New creates a Buffer and registers lifecycle hooks to
@@ -151,7 +142,7 @@ func New(
 			b.mu.Unlock()
-			b.writers.Go(b.flushLoop)
+			go b.flushLoop()
 			return nil
 		},
@@ -162,14 +153,7 @@ func New(
 			var err error
 			b.stopOnce.Do(func() {
 				// Closing done ends flushLoop after any flush it is
 				// running. Wait for it and for the writes Append
 				// started, which the process exiting would cut off;
 				// one that fails puts its reports back in buf, for the
 				// final flush to write.
 				close(b.done)
 				b.writers.Wait()
 				err = b.flushLocked()
 			})
@@ -183,12 +167,11 @@ func New(
 }
 // Append marshals v as a single JSON line and appends it to
-// the buffer. It stores nothing and returns an error if the last
+// the buffer. If the line would take usedBytes past maxBytes, the
-// write of a report file failed, until a write succeeds again. If the
+// oldest report files are deleted to make room; it stores nothing
-// line would take usedBytes past maxBytes, the oldest report files are
+// and returns ErrFull if that cannot make room. If the buffer
-// deleted to make room; it stores nothing and returns ErrFull if that
+// reaches the size threshold, it is drained and written to disk
-// cannot make room. If the buffer reaches the size threshold, it is
+// asynchronously.
 // drained and written to disk asynchronously.
 func (b *Buffer) Append(v any) error {
 	line, err := json.Marshal(v)
 	if err != nil {
@@ -199,13 +182,6 @@ func (b *Buffer) Append(v any) error {
 	b.mu.Lock()
 	if b.writeErr != nil {
 		err = fmt.Errorf("report files cannot be written: %w", b.writeErr)
 		b.mu.Unlock()
 		return err
 	}
 	b.deleteOldestFiles(lineBytes)
 	if b.usedBytes+lineBytes > b.maxBytes {
@@ -222,12 +198,12 @@ func (b *Buffer) Append(v any) error {
 		data := b.drainBuf()
 		b.mu.Unlock()
-		b.writers.Go(func() {
+		go func() {
 			writeErr := b.writeFile(data)
 			if writeErr != nil {
 				b.log.Error("flush reports failed", "error", writeErr)
 			}
-		})
+		}()
 		return nil
 	}
@@ -315,8 +291,7 @@ func (b *Buffer) drainBuf() []byte {
 }
 // writeFile writes data, reports drained from the buffer, to a new
-// timestamped zstd-compressed JSONL file in the data directory. If the
+// timestamped zstd-compressed JSONL file in the data directory.
 // write fails, it puts the reports back in the buffer.
 func (b *Buffer) writeFile(data []byte) error {
 	// The timestamp comes first, so the names sort by time; the number
 	// after it tells apart files named in the same millisecond.
@@ -325,27 +300,19 @@ func (b *Buffer) writeFile(data []byte) error {
 	size, err := b.createFile(filepath.Join(b.dataDir, name), data)
 	// The reports no longer wait to be written, so they stop counting
 	// at their uncompressed size. If the write failed they are lost;
 	// otherwise they count as the file, which from here on may be
 	// deleted to make room.
 	b.mu.Lock()
 	defer b.mu.Unlock()
-	b.writeErr = err
+	b.usedBytes -= int64(len(data))
 	if err != nil {
 		// The reports go back in front of any appended since, for the
 		// next flush to write, and still count at their uncompressed
 		// size.
 		appended := bytes.Clone(b.buf.Bytes())
 		b.buf.Reset()
 		b.buf.Write(data)
 		b.buf.Write(appended)
 		return err
 	}
 	// The reports no longer wait to be written, so they stop counting
 	// at their uncompressed size and count as the file, which from here
 	// on may be deleted to make room.
 	b.usedBytes -= int64(len(data))
 	b.usedBytes += size
 	b.filesBytes += size
@@ -5,8 +5,6 @@ import (
 	"errors"
 	"fmt"
 	"io/fs"
 	"net/http"
 	"net/http/httptest"
 	"os"
 	"path/filepath"
 	"slices"
@@ -15,13 +13,10 @@ import (
 	"sync"
 	"sync/atomic"
 	"testing"
 	"testing/synctest"
 	"time"
 	"sneak.berlin/go/netwatch/internal/config"
 	"sneak.berlin/go/netwatch/internal/globals"
 	"sneak.berlin/go/netwatch/internal/handlers"
 	"sneak.berlin/go/netwatch/internal/healthcheck"
 	"sneak.berlin/go/netwatch/internal/logger"
 	"sneak.berlin/go/netwatch/internal/reportbuf"
@@ -108,174 +103,9 @@ func TestFailedFinalFlushFailsStop(t *testing.T) {
 	}
 }
-// TestStopWaitsForPeriodicWrite holds open the write the periodic flush
+// startBuffer starts a Buffer through fx, as main does, with the
-// starts, then stops the app. The stop must wait until the write is
+// DATA_DIR and DATA_DIR_MAX_BYTES the calling test has set.
-// released, and the report must then be in its file. The test runs in
+func startBuffer(t *testing.T) *reportbuf.Buffer {
 // a synctest bubble, whose clock moves on at once when every goroutine
 // in it is waiting, so the periodic flush does not take a minute.
 func TestStopWaitsForPeriodicWrite(t *testing.T) {
 	dir := t.TempDir()
 	t.Setenv("DATA_DIR", dir)
 	synctest.Test(t, func(t *testing.T) {
 		buf, app := startApp(t)
 		created := make(chan struct{})
 		release := make(chan struct{})
 		buf.OnFileCreated(func(*os.File) {
 			created <- struct{}{}
 			<-release
 		})
 		err := buf.Append(map[string]string{"id": "periodic"})
 		if err != nil {
 			t.Fatalf("append report: %v", err)
 		}
 		<-created
 		err = stopWhileHeld(t, app, release)
 		if err != nil {
 			t.Fatalf("stop: %v", err)
 		}
 		files, err := readReportFiles(dir)
 		if err != nil {
 			t.Fatalf("read report files: %v", err)
 		}
 		want := `{"id":"periodic"}` + "\n"
 		if len(files) != 1 || files[0] != want {
 			t.Fatalf("report files = %q, want one holding %q", files, want)
 		}
 	})
 }
 // TestStopWaitsForWriteAtSizeThreshold holds open the write Append
 // starts when the buffer reaches FlushSizeThreshold, then stops the
 // app. The stop must wait until the write is released, and the reports
 // must then be in its file.
 func TestStopWaitsForWriteAtSizeThreshold(t *testing.T) {
 	dir := t.TempDir()
 	t.Setenv("DATA_DIR", dir)
 	synctest.Test(t, func(t *testing.T) {
 		buf, app := startApp(t)
 		created := make(chan struct{})
 		release := make(chan struct{})
 		buf.OnFileCreated(func(*os.File) {
 			created <- struct{}{}
 			<-release
 		})
 		appended := appendToThreshold(t, buf)
 		<-created
 		err := stopWhileHeld(t, app, release)
 		if err != nil {
 			t.Fatalf("stop: %v", err)
 		}
 		files, err := readReportFiles(dir)
 		if err != nil {
 			t.Fatalf("read report files: %v", err)
 		}
 		if len(files) != 1 || files[0] != appended {
 			t.Fatalf("%d report files, want one holding the %d bytes "+
 				"appended", len(files), len(appended))
 		}
 	})
 }
 // TestFinalFlushWritesReportsOfFailedWrite holds open the write the
 // periodic flush starts, makes it fail, then stops the app. The stop
 // waits for the write, which puts its report back in the buffer, and
 // the final flush writes it, so the stop succeeds with nothing lost.
 func TestFinalFlushWritesReportsOfFailedWrite(t *testing.T) {
 	dir := t.TempDir()
 	t.Setenv("DATA_DIR", dir)
 	synctest.Test(t, func(t *testing.T) {
 		buf, app := startApp(t)
 		created := make(chan *os.File)
 		release := make(chan struct{})
 		buf.OnFileCreated(func(f *os.File) {
 			created <- f
 			<-release
 		})
 		err := buf.Append(map[string]string{"id": "failed"})
 		if err != nil {
 			t.Fatalf("append report: %v", err)
 		}
 		held := <-created
 		// Closing the file under the write makes the write fail.
 		_ = held.Close()
 		// The final flush's write goes through.
 		buf.OnFileCreated(func(*os.File) {})
 		err = stopWhileHeld(t, app, release)
 		if err != nil {
 			t.Fatalf("stop: %v", err)
 		}
 		files, err := readReportFiles(dir)
 		if err != nil {
 			t.Fatalf("read report files: %v", err)
 		}
 		want := `{"id":"failed"}` + "\n"
 		if len(files) != 1 || files[0] != want {
 			t.Fatalf("report files = %q, want one holding %q", files, want)
 		}
 	})
 }
 // stopWhileHeld stops app while a report file write is held open until
 // release is closed, and fails the test if the stop returns before
 // that. It then closes release and returns the stop's error. It must
 // be called in a synctest bubble.
 func stopWhileHeld(
 	t *testing.T,
 	app *fxtest.App,
 	release chan<- struct{},
 ) error {
 	t.Helper()
 	stopped := make(chan error, 1)
 	go func() { stopped <- app.Stop(t.Context()) }()
 	// Wait returns once every other goroutine in the bubble is waiting:
 	// the stop has got as far as it can while the write is held.
 	synctest.Wait()
 	if len(stopped) != 0 {
 		t.Error("stop returned while a report file write was held open")
 	}
 	close(release)
 	return <-stopped
 }
 // startApp starts a Buffer through fx, as main does, with the DATA_DIR
 // and DATA_DIR_MAX_BYTES the calling test has set, and returns it with
 // the app, for tests that stop the app themselves.
 func startApp(t *testing.T) (*reportbuf.Buffer, *fxtest.App) {
 	t.Helper()
 	var buf *reportbuf.Buffer
@@ -293,16 +123,6 @@ func startApp(t *testing.T) (*reportbuf.Buffer, *fxtest.App) {
 	app.RequireStart()
 	t.Cleanup(app.RequireStop)
 	return buf, app
 }
 // startBuffer starts a Buffer through fx, as main does, with the
 // DATA_DIR and DATA_DIR_MAX_BYTES the calling test has set.
 func startBuffer(t *testing.T) *reportbuf.Buffer {
 	t.Helper()
 	buf, _ := startApp(t)
 	return buf
 }
@@ -687,17 +507,13 @@ func TestFileBeingWrittenIsNeverDeleted(t *testing.T) {
 	}
 }
-// TestFailedWriteKeepsItsReports makes a write fail once its file is
+// TestFailedWriteStopsCounting makes a write fail once its file is
-// created, under a cap with room for one report. The part of the file
+// created. Its reports are lost, so they stop counting, and the part of
-// written is removed, so it takes no room, and the report waits in
+// the file written is removed, so it takes no room.
-// memory for the next flush, which writes it. Once written, it counts
+func TestFailedWriteStopsCounting(t *testing.T) {
-// as its file and no longer as a report waiting, so the cap still holds
+	report := map[string]string{"id": "failed"}
 // exactly: one more report is taken and the one after it is refused.
 func TestFailedWriteKeepsItsReports(t *testing.T) {
 	report := map[string]string{"id": "kept"}
 	dir := t.TempDir()
-	t.Setenv("DATA_DIR", dir)
+	t.Setenv("DATA_DIR", t.TempDir())
 	t.Setenv("DATA_DIR_MAX_BYTES", strconv.Itoa(lineBytes(t, report)))
 	buf := startBuffer(t)
@@ -728,213 +544,10 @@ func TestFailedWriteKeepsItsReports(t *testing.T) {
 	// Writes from here on, the final one at stop included, succeed.
 	buf.OnFileCreated(func(*os.File) {})
 	err = buf.Flush()
 	if err != nil {
 		t.Fatalf("flush after the failed write: %v", err)
 	}
 	files, err := readReportFiles(dir)
 	if err != nil {
 		t.Fatalf("read report files: %v", err)
 	}
 	want := `{"id":"kept"}` + "\n"
 	if len(files) != 1 || files[0] != want {
 		t.Fatalf("report files = %q, want one holding %q", files, want)
 	}
 	// The file is deleted to make room if need be.
 	err = buf.Append(report)
 	if err != nil {
-		t.Fatalf("report after the written one: %v", err)
+		t.Fatalf("report after the failed write: %v", err)
 	}
 	err = buf.Append(report)
 	if !errors.Is(err, reportbuf.ErrFull) {
 		t.Fatalf("report past the cap: error = %v, want ErrFull", err)
 	}
 }
 // TestFailedWriteKeepsReportsAppendedDuringIt holds a write open,
 // appends a report while it is held, then makes the held write fail.
 // Both reports are kept, and the next flush writes them in the order
 // they were appended, the held write's first.
 func TestFailedWriteKeepsReportsAppendedDuringIt(t *testing.T) {
 	dir := t.TempDir()
 	t.Setenv("DATA_DIR", dir)
 	buf := startBuffer(t)
 	created := make(chan *os.File)
 	release := make(chan struct{})
 	buf.OnFileCreated(func(f *os.File) {
 		created <- f
 		<-release
 	})
 	err := buf.Append(map[string]string{"id": "held"})
 	if err != nil {
 		t.Fatalf("report of the held write: %v", err)
 	}
 	flushed := make(chan error)
 	go func() { flushed <- buf.Flush() }()
 	held := <-created
 	// Errorf, not Fatalf, until the held write is released, so that a
 	// failure here does not leave it held.
 	err = buf.Append(map[string]string{"id": "appended"})
 	if err != nil {
 		t.Errorf("report appended while the write was held: %v", err)
 	}
 	// Closing the file under the write makes the write fail.
 	_ = held.Close()
 	close(release)
 	err = <-flushed
 	if err == nil {
 		t.Fatal("held write succeeded, though its file was closed under it")
 	}
 	// Writes from here on, the final one at stop included, succeed.
 	buf.OnFileCreated(func(*os.File) {})
 	err = buf.Flush()
 	if err != nil {
 		t.Fatalf("flush after the failed write: %v", err)
 	}
 	files, err := readReportFiles(dir)
 	if err != nil {
 		t.Fatalf("read report files: %v", err)
 	}
 	want := `{"id":"held"}` + "\n" + `{"id":"appended"}` + "\n"
 	if len(files) != 1 || files[0] != want {
 		t.Fatalf("report files = %q, want one holding %q", files, want)
 	}
 }
 // TestReportsRefusedWhileWritesFail posts reports through the report
 // handler, as the page does. Once a write has failed, a report is
 // refused with 500 until a write succeeds again; the reports taken
 // before and after are stored, and the one refused is not.
 func TestReportsRefusedWhileWritesFail(t *testing.T) {
 	dir := t.TempDir()
 	t.Setenv("DATA_DIR", dir)
 	buf, h := startBufferWithHandlers(t)
 	if code := postReport(t, h, "before"); code != http.StatusOK {
 		t.Fatalf("report before the failed write: status = %d, want %d",
 			code, http.StatusOK)
 	}
 	// With the data directory gone, the write has nowhere to go.
 	err := os.Remove(dir)
 	if err != nil {
 		t.Fatalf("remove data dir: %v", err)
 	}
 	err = buf.Flush()
 	if err == nil {
 		t.Fatal("flush succeeded with the data directory gone")
 	}
 	if code := postReport(t, h, "refused"); code != http.StatusInternalServerError {
 		t.Fatalf("report after the failed write: status = %d, want %d",
 			code, http.StatusInternalServerError)
 	}
 	err = os.Mkdir(dir, 0o750)
 	if err != nil {
 		t.Fatalf("make data dir: %v", err)
 	}
 	err = buf.Flush()
 	if err != nil {
 		t.Fatalf("flush with the data directory back: %v", err)
 	}
 	if code := postReport(t, h, "after"); code != http.StatusOK {
 		t.Fatalf("report after a write succeeded: status = %d, want %d",
 			code, http.StatusOK)
 	}
 	err = buf.Flush()
 	if err != nil {
 		t.Fatalf("flush of the report after: %v", err)
 	}
 	files, err := readReportFiles(dir)
 	if err != nil {
 		t.Fatalf("read report files: %v", err)
 	}
 	stored := strings.Join(files, "")
 	for _, clientID := range []string{"before", "after"} {
 		if !strings.Contains(stored, `"clientId":"`+clientID+`"`) {
 			t.Errorf("report %q not stored", clientID)
 		}
 	}
 	if strings.Contains(stored, `"clientId":"refused"`) {
 		t.Error("refused report stored")
 	}
 }
 // startBufferWithHandlers starts a Buffer and the Handlers that store
 // reports in it through fx, as main does, with the DATA_DIR the calling
 // test has set.
 func startBufferWithHandlers(
 	t *testing.T,
 ) (*reportbuf.Buffer, *handlers.Handlers) {
 	t.Helper()
 	var (
 		buf *reportbuf.Buffer
 		h   *handlers.Handlers
 	)
 	app := fxtest.New(t,
 		fx.Provide(
 			config.New,
 			globals.New,
 			handlers.New,
 			healthcheck.New,
 			logger.New,
 			reportbuf.New,
 		),
 		fx.Populate(&buf, &h),
 	)
 	app.RequireStart()
 	t.Cleanup(app.RequireStop)
 	return buf, h
 }
 // postReport posts a report from clientID to the report handler, as the
 // page does, and returns the status of the answer.
 func postReport(t *testing.T, h *handlers.Handlers, clientID string) int {
 	t.Helper()
 	rec := httptest.NewRecorder()
 	req := httptest.NewRequestWithContext(t.Context(),
 		http.MethodPost, "/api/v1/reports",
 		strings.NewReader(`{"clientId":"`+clientID+`","hosts":[]}`),
 	)
 	req.Header.Set("Content-Type", "application/json")
 	h.HandleReport().ServeHTTP(rec, req)
 	return rec.Code
 }
 // TestConcurrentAppendsStopAtCap appends from many goroutines at once
@@ -1068,32 +681,6 @@ func TestFlushAtSizeThreshold(t *testing.T) {
 	buf := startBuffer(t)
 	appended := appendToThreshold(t, buf)
 	// Append writes the file in the background, so wait for it.
 	deadline := time.Now().Add(10 * time.Second)
 	for {
 		files, err := readReportFiles(dir)
 		if err == nil && len(files) == 1 && files[0] == appended {
 			return
 		}
 		if time.Now().After(deadline) {
 			t.Fatalf("%d report files (error: %v), want one holding the "+
 				"%d bytes appended", len(files), err, len(appended))
 		}
 		time.Sleep(10 * time.Millisecond)
 	}
 }
 // appendToThreshold appends reports to buf until it holds
 // FlushSizeThreshold bytes, so that the last append starts writing them
 // to a report file, and returns the lines appended.
 func appendToThreshold(t *testing.T, buf *reportbuf.Buffer) string {
 	t.Helper()
 	// Large reports, so the threshold takes a few hundred appends.
 	pad := strings.Repeat("a", 64<<10)
@@ -1116,7 +703,22 @@ func appendToThreshold(t *testing.T, buf *reportbuf.Buffer) string {
 		appended.WriteByte('\n')
 	}
-	return appended.String()
+	// Append writes the file in the background, so wait for it.
 	deadline := time.Now().Add(10 * time.Second)
 	for {
 		files, err := readReportFiles(dir)
 		if err == nil && len(files) == 1 && files[0] == appended.String() {
 			return
 		}
 		if time.Now().After(deadline) {
 			t.Fatalf("%d report files (error: %v), want one holding the "+
 				"%d bytes appended", len(files), err, appended.Len())
 		}
 		time.Sleep(10 * time.Millisecond)
 	}
 }
 // reportFilesBytes returns the total size of the report files in dir.
@@ -0,0 +1,11 @@
 package server
 // MaxRequestBodyBytes exposes the router-wide body limit to the
 // external tests.
 const MaxRequestBodyBytes = maxRequestBodyBytes
 // ListenAddr exposes the address the server listens on to the
 // external tests.
 func (s *Server) ListenAddr() string {
 	return s.newHTTPServer().Addr
 }
@@ -3,7 +3,6 @@ package server
 import (
 	"time"
 	sentryhttp "github.com/getsentry/sentry-go/http"
 	"github.com/go-chi/chi/v5"
 	"github.com/go-chi/chi/v5/middleware"
 	"github.com/prometheus/client_golang/prometheus"
@@ -18,12 +17,6 @@ const (
 	// can mount s.mw.MaxBodyBytes with a smaller value to lower
 	// its bound, but cannot raise it: this cap runs first.
 	maxRequestBodyBytes int64 = 1 << 20 // 1 MiB
 	// metricsRequestsPerMinute is how many requests to /metrics each
 	// client address may make a minute, whatever their credentials. A
 	// scraper polling every 2 seconds sends half of it, which httprate
 	// never refuses.
 	metricsRequestsPerMinute = 60
 )
 // SetupRoutes configures the chi router with middleware and
@@ -34,20 +27,11 @@ func (s *Server) SetupRoutes() {
 	s.router.Use(s.mw.Recoverer())
 	s.router.Use(middleware.RequestID)
 	s.router.Use(s.mw.Logging())
 	// s.mw.Recoverer would write a panic's 500 only after the panic has
 	// passed the request log, so it is written here, where the log sees it.
 	s.router.Use(s.mw.Write500OnPanic())
 	s.router.Use(s.mw.SecurityHeaders())
 	s.router.Use(s.mw.CORS(s.params.Config.CORSAllowedOrigins))
 	s.router.Use(s.mw.MaxBodyBytes(maxRequestBodyBytes))
 	s.router.Use(middleware.Timeout(requestTimeout))
 	// Sentry reports a panic, then panics again, so that it still
 	// reaches s.mw.Recoverer.
 	if s.params.Config.SentryDSN != "" {
 		s.router.Use(sentryhttp.New(sentryhttp.Options{Repanic: true}).Handle)
 	}
 	// The metrics go in a registry of this server's own, not in
 	// Prometheus' default one, which takes them only once per process.
 	registry := prometheus.NewRegistry()
@@ -67,8 +51,6 @@ func (s *Server) SetupRoutes() {
 		// other.
 		if s.params.Config.MetricsUsername != "" {
 			r.Use(s.mw.Metrics(registry))
 			// Likewise inside the metrics.
 			r.Use(s.mw.Write500OnPanic())
 		}
 		r.Get("/.well-known/healthcheck", s.h.HandleHealthCheck())
@@ -77,13 +59,9 @@ func (s *Server) SetupRoutes() {
 			Post("/api/v1/reports", s.h.HandleReport())
 	})
 	// The rate limit comes before the basic auth, so a client past it
 	// gets 429 and its password is not checked.
 	if s.params.Config.MetricsUsername != "" {
-		s.router.With(
+		s.router.With(s.mw.MetricsAuth()).
-			s.mw.RateLimit(metricsRequestsPerMinute),
+			Get("/metrics", promhttp.HandlerFor(
 			s.mw.MetricsAuth(),
 		).Get("/metrics", promhttp.HandlerFor(
 				registry, promhttp.HandlerOpts{},
 			).ServeHTTP)
 	}
@@ -1,12 +1,10 @@
 package server_test
 import (
 	"io"
 	"net/http"
 	"net/http/httptest"
 	"strings"
 	"testing"
 	"time"
 	"sneak.berlin/go/netwatch/internal/config"
 	"sneak.berlin/go/netwatch/internal/globals"
@@ -17,7 +15,6 @@ import (
 	"sneak.berlin/go/netwatch/internal/reportbuf"
 	"sneak.berlin/go/netwatch/internal/server"
 	"github.com/getsentry/sentry-go"
 	"go.uber.org/fx"
 	"go.uber.org/fx/fxtest"
 )
@@ -66,7 +63,6 @@ func TestReportsAreRateLimited(t *testing.T) {
 			http.MethodPost, "/api/v1/reports",
 			strings.NewReader(`{"clientId":"c1","hosts":[]}`),
 		)
 		req.Header.Set("Content-Type", "application/json")
 		srv.ServeHTTP(rec, req)
 		return rec.Code
@@ -189,50 +185,6 @@ func TestMetricsBehindBasicAuth(t *testing.T) {
 	}
 }
 // TestMetricsAreRateLimited: a client that has used up its /metrics
 // allowance on wrong passwords gets 429 even with the right one, which
 // is then not checked, while another client behind the same nginx
 // still gets in.
 func TestMetricsAreRateLimited(t *testing.T) {
 	t.Setenv("METRICS_USERNAME", "prometheus")
 	t.Setenv("METRICS_PASSWORD", "right")
 	// As in the container: nginx connects from loopback and names the
 	// client in X-Forwarded-For.
 	t.Setenv("TRUSTED_PROXIES", "127.0.0.1/32")
 	srv := newServer(t)
 	srv.SetupRoutes()
 	get := func(client, password string) int {
 		rec := httptest.NewRecorder()
 		req := httptest.NewRequestWithContext(t.Context(),
 			http.MethodGet, "/metrics", http.NoBody)
 		req.RemoteAddr = "127.0.0.1:40000"
 		req.Header.Set("X-Forwarded-For", client)
 		req.SetBasicAuth("prometheus", password)
 		srv.ServeHTTP(rec, req)
 		return rec.Code
 	}
 	for i := range server.MetricsRequestsPerMinute {
 		if code := get("203.0.113.7", "wrong"); code != http.StatusUnauthorized {
 			t.Fatalf("guess %d: status = %d, want %d",
 				i+1, code, http.StatusUnauthorized)
 		}
 	}
 	if code := get("203.0.113.7", "right"); code != http.StatusTooManyRequests {
 		t.Fatalf("right password past the limit: status = %d, want %d",
 			code, http.StatusTooManyRequests)
 	}
 	if code := get("203.0.113.8", "right"); code != http.StatusOK {
 		t.Fatalf("another client: status = %d, want %d",
 			code, http.StatusOK)
 	}
 }
 // TestMetricsInTwoServers: two servers in one process can both have
 // metrics on.
 func TestMetricsInTwoServers(t *testing.T) {
@@ -244,70 +196,6 @@ func TestMetricsInTwoServers(t *testing.T) {
 	}
 }
 // TestSentry: with SENTRY_DSN empty there is no Sentry client. With it
 // pointing at a local server standing in for Sentry, a panic in a
 // handler reaches that server, and the request still gets the 500 from
 // the panic recovery.
 func TestSentry(t *testing.T) {
 	const panicMessage = "handler panic for TestSentry"
 	// sentry.Init sets the client for the whole process; take it away
 	// again so that no other test reports to Sentry.
 	t.Cleanup(func() { sentry.CurrentHub().BindClient(nil) })
 	t.Setenv("SENTRY_DSN", "")
 	newServer(t)
 	if sentry.CurrentHub().Client() != nil {
 		t.Fatal("a Sentry client exists with SENTRY_DSN empty")
 	}
 	// The body of the first request the stand-in for Sentry receives.
 	received := make(chan string, 1)
 	sentryServer := httptest.NewServer(http.HandlerFunc(
 		func(_ http.ResponseWriter, r *http.Request) {
 			body, _ := io.ReadAll(r.Body)
 			select {
 			case received <- string(body):
 			default:
 			}
 		},
 	))
 	defer sentryServer.Close()
 	t.Setenv("SENTRY_DSN",
 		"http://key@"+sentryServer.Listener.Addr().String()+"/1")
 	srv := newServer(t)
 	srv.SetupRoutes()
 	srv.Router().Get("/panic", func(http.ResponseWriter, *http.Request) {
 		panic(panicMessage)
 	})
 	rec := httptest.NewRecorder()
 	req := httptest.NewRequestWithContext(t.Context(),
 		http.MethodGet, "/panic", http.NoBody)
 	srv.ServeHTTP(rec, req)
 	if rec.Code != http.StatusInternalServerError {
 		t.Fatalf("status = %d, want %d",
 			rec.Code, http.StatusInternalServerError)
 	}
 	// Sentry sends from a goroutine of its own.
 	select {
 	case body := <-received:
 		if !strings.Contains(body, panicMessage) {
 			t.Fatalf("the Sentry server received no report of the panic:\n%s",
 				body)
 		}
 	case <-time.After(5 * time.Second):
 		t.Fatal("nothing reached the Sentry server")
 	}
 }
 // TestHealthCheckRejectsOversizeBody sends the health check, which
 // never reads its body, a body one byte over the limit. Only the
 // router-wide body limit can reject it.
@@ -7,10 +7,8 @@ package server
 import (
 	"context"
 	"fmt"
 	"log/slog"
 	"net/http"
 	"time"
 	"sneak.berlin/go/netwatch/internal/config"
 	"sneak.berlin/go/netwatch/internal/globals"
@@ -18,15 +16,10 @@ import (
 	"sneak.berlin/go/netwatch/internal/logger"
 	"sneak.berlin/go/netwatch/internal/middleware"
 	"github.com/getsentry/sentry-go"
 	"github.com/go-chi/chi/v5"
 	"go.uber.org/fx"
 )
 // sentryFlushTimeout is how long shutdown waits for Sentry to send
 // what it still holds.
 const sentryFlushTimeout = 2 * time.Second
 // Params defines the dependencies for Server.
 type Params struct {
 	fx.In
@@ -63,11 +56,6 @@ func New(
 	s.log = params.Logger.Get()
 	s.shutdowner = params.Shutdowner
 	err := s.enableSentry()
 	if err != nil {
 		return nil, err
 	}
 	lc.Append(fx.Hook{
 		OnStart: func(_ context.Context) error {
 			// Build the router and http.Server synchronously
@@ -99,37 +87,10 @@ func (s *Server) ServeHTTP(
 	s.router.ServeHTTP(w, r)
 }
 // enableSentry sets Sentry up when SENTRY_DSN is set, so that
 // SetupRoutes can report panics to it. With SENTRY_DSN empty it does
 // nothing. A DSN Sentry refuses stops the start.
 func (s *Server) enableSentry() error {
 	if s.params.Config.SentryDSN == "" {
 		return nil
 	}
 	err := sentry.Init(sentry.ClientOptions{
 		Dsn:     s.params.Config.SentryDSN,
 		Release: s.params.Globals.Appname + "-" + s.params.Globals.Version,
 	})
 	if err != nil {
 		return fmt.Errorf("SENTRY_DSN: %w", err)
 	}
 	s.log.Info("sentry error reporting activated")
 	return nil
 }
 // shutdown gracefully stops the HTTP server within the
-// deadline of the context fx provides for OnStop, then gives
+// deadline of the context fx provides for OnStop.
 // Sentry, if set up, time to send what it still holds.
 func (s *Server) shutdown(ctx context.Context) error {
 	err := s.httpServer.Shutdown(ctx)
 	if s.params.Config.SentryDSN != "" {
 		sentry.Flush(sentryFlushTimeout)
 	}
 	if err != nil {
 		s.log.Error("server clean shutdown failed", "error", err)
@@ -0,0 +1,21 @@
 #!/bin/sh
 # script/build: compile the static netwatch-server binary into the
 # backend project root, with its version stamped in.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    # VERSION comes from the environment (the root Dockerfile passes its
    # ARG VERSION in). Unset or empty, it is git describe, or "dev" where
    # there is no git or no repository history.
    version="${VERSION:-$(git describe --always --dirty 2>/dev/null || echo dev)}"
    CGO_ENABLED=0 go build -trimpath \
        -ldflags "-s -w -X main.Version=$version" \
        -o netwatch-server ./cmd/netwatch-server/
 }
 main "$@"
@@ -0,0 +1,12 @@
 #!/bin/sh
 # script/clean: remove build artifacts.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    rm -f netwatch-server
 }
 main "$@"
@@ -0,0 +1,12 @@
 #!/bin/sh
 # script/fmt: format the Go sources (writes).
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    go fmt ./...
 }
 main "$@"
@@ -0,0 +1,18 @@
 #!/bin/sh
 # script/fmt-check: check Go formatting (read-only). Same scope as
 # script/fmt, but fails instead of writing.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    unformatted="$(gofmt -l .)"
    if [ -n "$unformatted" ]; then
        echo "Files not formatted:" >&2
        echo "$unformatted" >&2
        exit 1
    fi
 }
 main "$@"
@@ -0,0 +1,50 @@
 #!/bin/sh
 # script/lint: run golangci-lint over the backend. This runs inside the
 # lint stage of the root Dockerfile, whose digest-pinned golangci-lint
 # image provides the linter; nothing installs golangci-lint on the host.
 # From a checkout, run `make lint` at the repo root, which builds that
 # stage.
 #
 # .golangci.yml is standardized org-wide and must never be edited here
 # (REPO_POLICIES.md). Its last silent drift replaced the v2 schema with
 # v1 keys, which left every threshold in the file inert while the build
 # stayed green. So the file is first checked against the canonical
 # copy's sha256: a local comparison, no network, nothing unpinned.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 # The sha256 of the org standard .golangci.yml. When that file changes in
 # sneak/prompts and is copied here again, this changes with it.
 GOLANGCI_CONFIG_SHA256="a79b63a254602a5318db5d0e9a06bc71b84bf0c1d896305229d8bfed1d1b1776"
 main() {
    cd "$ROOT"
    if [ ! -f .golangci.yml ]; then
        echo "backend/.golangci.yml is missing. Copy the org standard verbatim" >&2
        echo "from https://git.eeqj.de/sneak/prompts/raw/branch/main/.golangci.yml" >&2
        exit 1
    fi
    actual="$(sha256sum .golangci.yml | cut -d' ' -f1)"
    if [ -z "$actual" ]; then
        echo "sha256sum is missing or printed no hash, so" >&2
        echo "backend/.golangci.yml could not be checked." >&2
        exit 1
    fi
    if [ "$actual" != "$GOLANGCI_CONFIG_SHA256" ]; then
        echo "backend/.golangci.yml does not match GOLANGCI_CONFIG_SHA256" >&2
        echo "in backend/script/lint." >&2
        echo "  expected $GOLANGCI_CONFIG_SHA256" >&2
        echo "  actual   $actual" >&2
        echo "Compare it with the org standard," >&2
        echo "https://git.eeqj.de/sneak/prompts/raw/branch/main/.golangci.yml" >&2
        echo "- If they differ, it was edited here: restore the org standard" >&2
        echo "  verbatim. Do not edit it." >&2
        echo "- If they are the same, the org standard changed: set" >&2
        echo "  GOLANGCI_CONFIG_SHA256 in backend/script/lint to the actual hash." >&2
        exit 1
    fi
    golangci-lint run ./...
 }
 main "$@"
@@ -1,6 +1,5 @@
 #!/bin/sh
-# script/run: build with script/build, then run netwatch-server locally
+# script/run: build and run netwatch-server locally.
 # with its defaults: port 8080, reports stored in data/reports.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
@@ -0,0 +1,20 @@
 #!/bin/sh
 # script/test: run the backend test suite with the race detector and
 # coverage. Go's own -timeout bounds the tests and not their compile,
 # so a cold build cache cannot fail it. The race detector needs cgo,
 # and so a C compiler. If the tests fail, they run again with -v for
 # the details, and the script fails even if that run passes.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    go test -timeout 30s -race -cover ./... || {
        echo "--- Rerunning with -v for details ---"
        go test -timeout 30s -race -v ./...
        exit 1
    }
 }
 main "$@"
@@ -34,11 +34,9 @@ fi
 # TRUSTED_PROXIES names the reverse proxies in front of the container,
 # as IP addresses or CIDRs separated by commas. nginx takes the client
-# address from X-Forwarded-For only on a request from one of them.
+# address from X-Forwarded-For only on a request from one of them, so
-# Unset, they are the three private ranges of RFC 1918; set and empty,
+# unset or empty, it trusts no one. nginx.conf includes the file written
-# nginx trusts no one, which is why the default below follows - and not
+# here, one set_real_ip_from line per entry.
 # :-. configs/nginx.conf includes the file written here, one
 # set_real_ip_from line per entry.
 #
 # nginx looks up an entry it cannot read as an address as a hostname,
 # and trusts what it finds (1.2.3 is found as 1.2.0.3). So each entry
@@ -47,7 +45,7 @@ fi
 # TRUSTED_PROXIES. Its error, naming the CIDR, is dropped for the one
 # below, naming the entry as written. set -f keeps a * in an entry from
 # becoming a list of file names.
-TRUSTED_PROXIES="${TRUSTED_PROXIES-10.0.0.0/8,172.16.0.0/12,192.168.0.0/16}"
+TRUSTED_PROXIES="${TRUSTED_PROXIES:-}"
 set -f
 for proxy in $(printf '%s' "$TRUSTED_PROXIES" | tr ',' ' '); do
    case "$proxy" in
@@ -78,10 +76,10 @@ stop_requested=""
 trap 'stop_requested=yes' TERM INT
 # netwatch-server runs as the netwatch user and listens on loopback
-# only, on a port other than the public one; configs/nginx.conf proxies
+# only, on a port other than the public one; nginx.conf proxies to this
-# to this address. Its only client is nginx, so it takes the client
+# address. Its only client is nginx, so it takes the client address
-# address nginx passes on from 127.0.0.1 alone, whatever TRUSTED_PROXIES
+# nginx passes on from 127.0.0.1 alone, whatever TRUSTED_PROXIES the
-# the container has. The netwatch user has no login shell, hence -s
+# container has. The netwatch user has no login shell, hence -s
 # /bin/sh. busybox su replaces itself with the command instead of
 # staying on as its parent, so $! is the server's own PID.
 BIND_ADDRESS=127.0.0.1 PORT=8081 TRUSTED_PROXIES=127.0.0.1/32 \
@@ -90,8 +88,8 @@ backend=$!
 # nginx starts through the nginx image's own entrypoint, which applies
 # the image's start-up configuration and then replaces itself with
-# nginx. Part of that start-up configuration renders configs/nginx.conf
+# nginx. Part of that start-up configuration renders nginx.conf into
-# into conf.d with nginx listening on PORT. NGINX_ENVSUBST_FILTER limits
+# conf.d with nginx listening on PORT. NGINX_ENVSUBST_FILTER limits
 # that rendering to PORT: a variable nginx itself uses, such as $uri,
 # would otherwise be replaced by an environment variable of the same
 # name.
@@ -1,5 +1,5 @@
 // eslint's recommended rules over every JavaScript file in the repo.
-// make lint runs it, in the frontend-lint stage of Dockerfile.
+// script/frontend-lint runs it, in the frontend-lint stage of Dockerfile.
 import js from "@eslint/js";
 import globals from "globals";
 import { defineConfig } from "eslint/config";
@@ -8,7 +8,7 @@ export default defineConfig([
    js.configs.recommended,
    {
        // The page, and facts.js, which the viewport test runs inside it.
-        files: ["web/src/**/*.js", "test/viewport/facts.js"],
+        files: ["src/**/*.js", "test/viewport/facts.js"],
        languageOptions: {
            globals: {
                ...globals.browser,
@@ -9,7 +9,7 @@
            type="image/svg+xml"
            href="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 100 100'><text y='.9em' font-size='90'>📡</text></svg>"
        />
-        <!-- Linked here, not imported by web/src/main.js, so the unit tests can
+        <!-- Linked here, not imported by src/main.js, so the unit tests can
             import that module in Node, which cannot import CSS. -->
        <link rel="stylesheet" href="/src/styles.css" />
    </head>
@@ -1,23 +0,0 @@
 package server
 import "github.com/go-chi/chi/v5"
 // Router exposes the router to the external tests, which add routes
 // of their own to it after SetupRoutes.
 func (s *Server) Router() *chi.Mux {
 	return s.router
 }
 // MaxRequestBodyBytes exposes the router-wide body limit to the
 // external tests.
 const MaxRequestBodyBytes = maxRequestBodyBytes
 // MetricsRequestsPerMinute exposes the /metrics rate limit to the
 // external tests.
 const MetricsRequestsPerMinute = metricsRequestsPerMinute
 // ListenAddr exposes the address the server listens on to the
 // external tests.
 func (s *Server) ListenAddr() string {
 	return s.newHTTPServer().Addr
 }
@@ -18,10 +18,9 @@ server {
    # The client address comes from X-Forwarded-For only on a request
    # from the reverse proxies in TRUSTED_PROXIES: bin/entrypoint.sh
-    # writes one set_real_ip_from line for each into this file. Unset,
+    # writes one set_real_ip_from line for each into this file, and
-    # TRUSTED_PROXIES is the three private ranges of RFC 1918; set and
+    # leaves it empty when TRUSTED_PROXIES is unset, so that by default
-    # empty, the file is empty and the client address is the one each
+    # the client address is the one each request comes from.
    # request comes from.
    include /etc/nginx/trusted-proxies.conf;
    real_ip_header X-Forwarded-For;
    real_ip_recursive on;
@@ -54,7 +53,7 @@ server {
    # netwatch-server sets the same security headers on its own
    # responses. Its copies are dropped so that each header goes out
-    # once, as configs/security-headers.conf sets it.
+    # once, as security-headers.conf sets it.
    proxy_hide_header Strict-Transport-Security;
    proxy_hide_header Content-Security-Policy;
    proxy_hide_header X-Frame-Options;
@@ -6,18 +6,19 @@
 # used directly if it is at least NODE_MIN_VERSION; otherwise it is
 # installed at a pinned version via nvm (installing nvm itself first,
 # from a hash-verified release archive, never curl | sh). Go, with its
-# gofmt, is used directly only if it is exactly GO_VERSION; otherwise
+# gofmt, is used directly if it is at least the version backend/go.mod
-# that release is installed from its hash-verified archive.
+# asks for; otherwise the pinned Go release is installed from its
 # hash-verified archive.
 #
 # What this script installs outside the system package manager lives
 # under $HOME and is linked into ~/.local/bin, where make and the git
 # hook find it once that directory is on PATH. Nothing in ~/.local/bin
 # that this script did not create is ever replaced.
 #
-# golangci-lint is not installed: make lint runs it in Docker, as make
+# golangci-lint is not installed: make lint runs it in Docker, which
-# test runs the tests, and this script does not install Docker either.
+# this script does not install either.
 #
-# Unlike the org model: Go for the backend, a newer node for eslint.
+# Unlike the org model: Go and gcc for backend/, a newer node for eslint.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
@@ -185,30 +186,20 @@ ensure_yarn() {
 }
 # go_ok: the go on PATH has its gofmt beside it (a Go release ships the
-# two together) and go version reports exactly GO_VERSION, compared over
+# two together) and is at least the version backend/go.mod asks for.
-# the whole version, not a prefix of it. A go that fails or prints
+# GOTOOLCHAIN=local makes an older go fail here instead of fetching a
-# anything else does not pass. GOTOOLCHAIN=local makes go report itself
+# newer toolchain for itself.
 # rather than a toolchain it would fetch.
 go_ok() {
    if missing go; then return 1; fi
    [ -x "$(dirname "$(command -v go)")/gofmt" ] || return 1
-    version="$(GOTOOLCHAIN=local go version 2>/dev/null)" || return 1
+    (cd "$ROOT/backend" && GOTOOLCHAIN=local go list -m >/dev/null 2>&1)
    case "$version" in
        "go version go$GO_VERSION "*) return 0 ;;
        *) return 1 ;;
    esac
 }
 # ensure_go: unless go_ok, install GO_VERSION and link its go and gofmt.
 # They are linked on every run that needs them, so a deleted link is put
 # back, and the archive is unpacked again if either binary is missing.
 # Afterwards go is looked up through PATH again and must pass go_ok, so
 # another go that hides the link stops bootstrap.
 ensure_go() {
-    if go_ok; then
+    if go_ok; then return 0; fi
        echo "bootstrap: using $(GOTOOLCHAIN=local go version)"
        return 0
    fi
    go_dir="$TOOLCHAIN/go-$GO_VERSION"
    if [ ! -x "$go_dir/bin/go" ] || [ ! -x "$go_dir/bin/gofmt" ]; then
        # sha256 of each archive, from https://go.dev/dl/?mode=json
@@ -249,13 +240,6 @@ ensure_go() {
    fi
    link_bin "$go_dir/bin/go" go
    link_bin "$go_dir/bin/gofmt" gofmt
    hash -r
    if ! go_ok; then
        echo "bootstrap: after installing go $GO_VERSION in $go_dir," >&2
        echo "  the go on PATH, $(command -v go), is not it or has no gofmt" >&2
        exit 1
    fi
    echo "bootstrap: installed $(GOTOOLCHAIN=local go version)"
 }
 main() {
@@ -272,17 +256,23 @@ main() {
    if missing make; then pkg_install gnumake make make make; fi
    if missing git; then pkg_install git git git git; fi
    # The race detector in make test needs cgo, which Go turns on only
    # when it finds its C compiler, gcc on Linux. apt and apk ship the C
    # library headers apart from gcc.
    if missing gcc; then
        pkg_install gcc "gcc libc6-dev" gcc "gcc musl-dev"
    fi
    ensure_node
    ensure_yarn
    yarn install --frozen-lockfile
    ensure_go
-    go mod download
+    (cd "$ROOT/backend" && go mod download)
    if missing docker; then
-        echo "bootstrap: docker not found; make test and make lint, and so" >&2
+        echo "bootstrap: docker not found; make lint, and so make check" >&2
-        echo "  make check and the pre-commit hook, need it" >&2
+        echo "  and the pre-commit hook, need it to run the linters" >&2
    fi
    if [ -n "$path_hint" ] && [ -d "$BIN_DIR" ]; then
        echo "bootstrap: add $BIN_DIR to the front of your PATH, e.g." >&2
@@ -1,7 +1,6 @@
 #!/bin/sh
-# script/build: build the frontend for production into dist/, then the
+# script/build: build the frontend for production into dist/. The Go
-# static netwatch-server binary into the repo root, with its version
+# backend is built by backend/script/build.
 # stamped in.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
@@ -9,14 +8,6 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    yarn build
    # VERSION comes from the environment. Unset or empty, it is git
    # describe, or "dev" where there is no git or no repository history.
    version="${VERSION:-$(git describe --always --dirty 2>/dev/null || echo dev)}"
    CGO_ENABLED=0 go build -trimpath \
        -ldflags "-s -w -X main.Version=$version" \
        -o netwatch-server ./cmd/netwatch-server/
 }
 main "$@"
@@ -1,8 +1,6 @@
 #!/bin/sh
 # script/check: run all checks (test, lint, fmt-check). Our own
-# extension to scripts-to-rule-them-all. test and lint are Docker
+# extension to scripts-to-rule-them-all. Must not modify any files.
 # phases; fmt-check is native, because a formatter writes the working
 # tree. Must not modify any files.
 set -eu
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -1,13 +0,0 @@
 #!/bin/sh
 # script/clean: remove what script/build writes: dist/ and the
 # netwatch-server binary.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    rm -rf dist netwatch-server
 }
 main "$@"
@@ -1,20 +1,15 @@
 #!/bin/sh
-# script/fmt: format the whole repo (writes): prettier over the
+# script/fmt: format the whole repo (writes): prettier over everything
-# JavaScript, CSS, HTML and Markdown, then gofmt over the Go sources in
+# it understands, then gofmt over the Go backend.
-# cmd/ and internal/; not over the whole root, as node_modules/ holds
+# The org model formats only markdown; this repo also has JS and Go.
 # .go files of its own. The org model formats only markdown; this repo
 # also has JS and Go.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    # script/bootstrap links the node, yarn and gofmt it installs into
    # ~/.local/bin, which the shell that called it may not have on PATH.
    PATH="$HOME/.local/bin:$PATH"
    "$ROOT/script/frontend-fmt"
-    gofmt -l -w cmd internal
+    "$ROOT/backend/script/fmt"
 }
 main "$@"
@@ -8,16 +8,8 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    # script/bootstrap links the node, yarn and gofmt it installs into
    # ~/.local/bin, which the shell that called it may not have on PATH.
    PATH="$HOME/.local/bin:$PATH"
    "$ROOT/script/frontend-fmt-check"
-    unformatted="$(gofmt -l cmd internal)"
+    "$ROOT/backend/script/fmt-check"
    if [ -n "$unformatted" ]; then
        echo "Go files not formatted:" >&2
        echo "$unformatted" >&2
        exit 1
    fi
 }
 main "$@"
@@ -0,0 +1,18 @@
 #!/bin/sh
 # script/frontend-check: run the frontend tests and format check only.
 # This exists for the frontend stage of Dockerfile, a node image with
 # neither Go nor Docker; the Dockerfile's frontend-lint stage runs the
 # frontend linter, and its lint and builder stages gate the backend.
 # Everywhere else, use script/check, which covers the whole repo. Must
 # not modify any files.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    "$ROOT/script/frontend-test"
    "$ROOT/script/frontend-fmt-check"
 }
 main "$@"
@@ -1,16 +1,14 @@
 #!/bin/sh
-# script/frontend-fmt: format the JavaScript, CSS, HTML and Markdown,
+# script/frontend-fmt: format the frontend and every other file prettier
-# repo-wide (writes). Those are the languages REPO_POLICIES.md gives
+# understands, repo-wide (writes), the markdown in backend/ included.
-# prettier; the YAML is left alone, as .golangci.yml must stay the org
+# Prettier does not read Go; backend/script/fmt formats the Go sources.
 # standard byte for byte. Prettier does not read Go; script/fmt runs
 # gofmt over the Go sources.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
-    yarn prettier --write '**/*.{js,css,html,md}'
+    yarn prettier --write .
 }
 main "$@"
@@ -7,7 +7,7 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
-    yarn prettier --check '**/*.{js,css,html,md}'
+    yarn prettier --check .
 }
 main "$@"
@@ -0,0 +1,15 @@
 #!/bin/sh
 # script/frontend-lint: run eslint over the frontend. This runs inside
 # the frontend-lint stage of Dockerfile, the digest-pinned node image
 # with the packages from yarn.lock. From a checkout, run `make lint`,
 # which builds that stage.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
    yarn eslint .
 }
 main "$@"
@@ -1,11 +1,9 @@
 #!/bin/sh
-# script/frontend-test: run the frontend test suite on the host: the
+# script/frontend-test: run the frontend test suite: the unit tests in
-# unit tests in test/unit/, through the test script in package.json,
+# test/unit/, through the test script in package.json, then the
-# then the production build, which fails on broken code. make test runs
+# production build, which fails on broken code. The tests print a dot
-# the same tests in the frontend-test stage of Dockerfile, and the image
+# each; if any fails, they run again with every test listed, and the
-# build runs the production build in its frontend-build stage. The tests
+# script fails even if that run passes. NODE_OPTIONS chooses the
 # print a dot each; if any fails, they run again with every test listed,
 # and the script fails even if that run passes. NODE_OPTIONS chooses the
 # reporter because yarn adds its arguments after the test files, where
 # node would take a reporter option for one more file.
 set -eu
@@ -14,9 +12,9 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
    cd "$ROOT"
-    NODE_OPTIONS=--test-reporter=dot timeout 90 yarn --silent run test || {
+    NODE_OPTIONS=--test-reporter=dot timeout 30 yarn --silent run test || {
        echo "--- Rerunning with every test listed for details ---"
-        NODE_OPTIONS=--test-reporter=spec timeout 90 yarn --silent run test
+        NODE_OPTIONS=--test-reporter=spec timeout 30 yarn --silent run test
        exit 1
    }
    timeout 30 yarn build
@@ -2,11 +2,11 @@
 # script/frontend-viewport-test: verify the responsive layout of the built
 # frontend in a real browser engine.
 #
-# Builds dist/, serves it with the same nginx image and the same
+# Builds dist/, serves it with the same nginx image and the same nginx.conf
-# configs/nginx.conf the shipping container uses, points a containerised
+# the shipping container uses, points a containerised headless Chrome at it
-# headless Chrome at it over CDP, and asserts on computed layout at every
+# over CDP, and asserts on computed layout at every viewport width derived
-# viewport width derived from the app's own CSS. See
+# from the app's own CSS. See test/viewport/README.md for what this covers
-# test/viewport/README.md for what this covers and what it cannot.
+# and what it cannot.
 #
 # Deliberately not part of script/check: it needs Docker and takes far
 # longer than the 20s budget make test has to stay inside.
@@ -61,18 +61,18 @@ main() {
    # host.
    docker network create --internal "$NETWORK" > /dev/null
-    # configs/nginx.conf is a template: the image renders it over its own
+    # nginx.conf is a template: the image renders it over its own
    # default.conf, with the same port and limit bin/entrypoint.sh uses.
    # The empty file it includes trusts no proxy, as bin/entrypoint.sh
-    # writes it when TRUSTED_PROXIES is set and empty. It also includes
+    # writes it when TRUSTED_PROXIES is unset. nginx.conf also includes
    # the security headers, so the page runs under the shipped policy.
    docker run -d --rm --name "$SERVER" \
        --network "$NETWORK" --network-alias netwatch \
        -e PORT=8080 -e NGINX_ENVSUBST_FILTER='^PORT$' \
        -v "$ROOT/dist:/usr/share/nginx/html:ro" \
-        -v "$ROOT/configs/nginx.conf:/etc/nginx/templates/default.conf.template:ro" \
+        -v "$ROOT/nginx.conf:/etc/nginx/templates/default.conf.template:ro" \
        -v /dev/null:/etc/nginx/trusted-proxies.conf:ro \
-        -v "$ROOT/configs/security-headers.conf:/etc/nginx/security-headers.conf:ro" \
+        -v "$ROOT/security-headers.conf:/etc/nginx/security-headers.conf:ro" \
        "$SERVER_IMAGE" > /dev/null
    # The image's own entrypoint already exposes CDP on 9222 and passes
@@ -1,23 +1,23 @@
 #!/bin/sh
-# script/lint: run the linter. Linting is a phase of the Dockerfile and
+# script/lint: lint the whole repo: eslint over the frontend, then the Go
-# this builds that phase alone; the linter is never installed or run on
+# linter over backend/.
 # a developer host, where a shared result cache and a host-global lock
 # make its answer untrustworthy.
 #
-# The phase is not the last stage in the file, so it is built only when
+# No linter runs on the host: this builds the frontend-lint and lint
-# --target names it. --no-cache because a cached lint layer is a lint
+# stages of Dockerfile, the digest-pinned node and golangci-lint images.
-# that did not run. The tag makes each build replace the previous image
+# The first runs eslint; the second runs the backend's fmt-check and
-# instead of leaving a dangling one behind.
+# lint targets. --no-cache makes each linter really run every time
 # rather than reuse an earlier result, and each stage is built for its
 # checks alone, so no image is kept.
 set -eu
-SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
+ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 main() {
    cd "$ROOT"
-    docker build --no-cache \
+    timeout 300 docker build --no-cache --target frontend-lint \
-        --target lint \
+        --output type=cacheonly .
-        -t "$("$SCRIPT_DIR/projectname")-lint" .
+    timeout 300 docker build --no-cache --target lint \
        --output type=cacheonly .
 }
 main "$@"
@@ -1,19 +1,17 @@
 #!/bin/sh
-# script/test: run the test suite. Testing is a phase of the Dockerfile
+# script/test: run the test suite for the whole repo: the frontend at
-# and this builds that phase alone, on the same terms as script/lint:
+# the repo root, then the Go backend in backend/. Each half has its own
-# --target because a phase that is not the last stage is built only when
+# 30-second limit, and there is none around both: from a cold Go build
-# named, --no-cache because a cached test layer is a test that did not
+# cache, compiling the backend's tests with the race detector can take
-# run, and a tag so each build replaces the previous image.
+# 30 seconds on its own, and Go's -timeout leaves the compile out.
 set -eu
-SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
+ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 main() {
    cd "$ROOT"
-    docker build --no-cache \
+    script/frontend-test
-        --target test \
+    backend/script/test
        -t "$("$SCRIPT_DIR/projectname")-test" .
 }
 main "$@"
@@ -1,12 +1,12 @@
 #!/bin/sh
-# script/tidy: run go mod tidy, which adds the modules the Go sources
+# script/tidy: run go mod tidy in backend/, which adds the modules the
-# import, drops those they no longer do, and updates go.sum.
+# Go sources import, drops those they no longer do, and updates go.sum.
 set -eu
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 main() {
-    cd "$ROOT"
+    cd "$ROOT/backend"
    go mod tidy
 }
@@ -1,5 +1,5 @@
 # The security headers REPO_POLICIES.md requires on every response.
-# configs/nginx.conf includes this file, which Dockerfile copies to
+# nginx.conf includes this file, which Dockerfile copies to
 # /etc/nginx/security-headers.conf. always sends each header on error
 # responses too.
@@ -8,11 +8,11 @@ add_header Strict-Transport-Security "max-age=31536000; includeSubDomains" alway
 # Scripts and styles load only from the page's own origin. Inline ones
 # are blocked, style attributes in markup included, so style elements
 # through classes or element.style. data: images are for the favicon
-# in web/index.html. connect-src is * because the browser checks each
+# in index.html. connect-src is * because the browser checks each probe in
-# probe in web/src/main.js against it, and also every redirect the probe
+# src/main.js against it, and also every redirect the probe follows,
-# follows, and several of those hosts redirect to others; a list of
+# and several of those hosts redirect to others; a list of hosts here
-# hosts here would block those probes. It also covers the reports the
+# would block those probes. It also covers the reports the page sends
-# page sends to its own origin.
+# to its own origin.
 add_header Content-Security-Policy "default-src 'self'; connect-src *; img-src 'self' data:; object-src 'none'; base-uri 'none'; form-action 'none'; frame-ancestors 'none'" always;
 add_header X-Frame-Options DENY always;
@@ -30,39 +30,6 @@ export const CONFIG = {
        return [0, 1, 2, 3, 4, 5].map((i) => Math.round((d * i) / 5));
    },
    canvasHeight: 96,
    // A latency figure and its sparkline take the color of the first entry
    // whose limit, in ms, the latency is below.
    latencyColors: [
        { below: 50, hex: "#22c55e", className: "text-green-500" },
        { below: 100, hex: "#84cc16", className: "text-lime-500" },
        { below: 200, hex: "#eab308", className: "text-yellow-500" },
        { below: 500, hex: "#f97316", className: "text-orange-500" },
        { below: Infinity, hex: "#ef4444", className: "text-red-500" },
    ],
    // The health is offline when more than offlineTimeouts WAN hosts timed
    // out or were unreachable and at most offlineReachable answered;
    // otherwise degraded when more than degradedTimeouts timed out or were
    // unreachable; otherwise slow when more than slowHosts answered after
    // more than slowLatency ms.
    offlineTimeouts: 10,
    offlineReachable: 4,
    degradedTimeouts: 4,
    slowHosts: 3,
    slowLatency: 1000,
    // The debug log keeps its last maxLogEntries lines.
    maxLogEntries: 1000,
    // A gateway candidate that has not answered after gatewayTimeout ms is
    // passed over.
    gatewayTimeout: 1500,
    // When no WAN host answers, the recovery probe checks recoveryProbeHosts
    // random ones every recoveryProbeInterval ms.
    recoveryProbeHosts: 4,
    recoveryProbeInterval: 500,
    // The rows are sorted after the first round that is not discarded, then
    // every roundsPerSort rounds.
    roundsPerSort: 10,
    // The sparklines are sized and drawn again resizeDelay ms after start.
    resizeDelay: 100,
 };
 // WAN endpoints to monitor. These are used for the aggregate health/stats
@@ -144,15 +111,12 @@ const WAN_HOSTS = [
 const debugLog = [];
-export const log = (() => {
+const log = (() => {
    function append(level, message) {
-        const entry = { timestamp: new Date(), level, message };
+        debugLog.push({ timestamp: new Date(), level, message });
-        debugLog.push(entry);
+        if (debugLog.length > 1000) debugLog.splice(0, debugLog.length - 1000);
        if (debugLog.length > CONFIG.maxLogEntries)
            debugLog.splice(0, debugLog.length - CONFIG.maxLogEntries);
        const panel = document.getElementById("debug-panel");
-        if (panel && !panel.classList.contains("hidden"))
+        if (panel && !panel.classList.contains("hidden")) renderDebugLog();
            appendDebugLogLine(entry);
    }
    return {
        error: (msg) => append("error", msg),
@@ -210,10 +174,7 @@ async function detectGateway() {
        const result = await Promise.any(
            GATEWAY_CANDIDATES.map(async (url) => {
                const controller = new AbortController();
-                const timeoutId = setTimeout(
+                const timeoutId = setTimeout(() => controller.abort(), 1500);
                    () => controller.abort(),
                    CONFIG.gatewayTimeout,
                );
                try {
                    await fetch(url, {
                        method: "GET",
@@ -238,35 +199,6 @@ async function detectGateway() {
 // --- App State ---------------------------------------------------------------
 // The min, max, median and average of latencies, a list of numbers, or all
 // null when it is empty. The median of an even count is the mean of the
 // middle two; it and the average are rounded.
 function latencyStats(latencies) {
    if (latencies.length === 0)
        return { min: null, max: null, med: null, avg: null };
    const sorted = [...latencies].sort((a, b) => a - b);
    const mid = Math.floor(sorted.length / 2);
    return {
        min: sorted[0],
        max: sorted[sorted.length - 1],
        med:
            sorted.length % 2
                ? sorted[mid]
                : Math.round((sorted[mid - 1] + sorted[mid]) / 2),
        avg: Math.round(
            latencies.reduce((a, b) => a + b, 0) / latencies.length,
        ),
    };
 }
 // The status a check's error gives its host: "error" for a timeout,
 // "offline" for any other error, and "online" for none.
 function statusFromError(error) {
    if (error === "timeout") return "error";
    if (error) return "offline";
    return "online";
 }
 export class HostState {
    constructor(host, pinned = false) {
        this.name = host.name;
@@ -275,15 +207,9 @@ export class HostState {
        // error }, or a round skipped while paused, { timestamp,
        // latency: null, paused: true }.
        this.history = [];
        // The last 100 (CONFIG.maxHistoryPoints) entries the interval menu
        // has cleared from the history, which the reporter may not have
        // sent yet.
        this.clearedHistory = [];
        this.lastLatency = null;
        this.status = "pending"; // 'online' | 'offline' | 'error' | 'pending'
        this.pinned = pinned;
        // How many recorded checks in a row have failed, up to the last one.
        this.consecutiveFailures = 0;
    }
    pushSample(timestamp, result) {
@@ -294,10 +220,9 @@ export class HostState {
        });
        this._trim();
        this.lastLatency = result.latency;
-        this.status = statusFromError(result.error);
+        if (result.error === "timeout") this.status = "error";
-        this.consecutiveFailures = result.error
+        else if (result.error) this.status = "offline";
-            ? this.consecutiveFailures + 1
+        else this.status = "online";
            : 0;
    }
    pushPaused(timestamp) {
@@ -305,26 +230,36 @@ export class HostState {
        this._trim();
    }
-    // What the interval menu does to each host: the history starts over and
+    averageLatency() {
-    // the host waits for its next check. The history's entries move to
+        const valid = this.history.filter((p) => p.latency !== null);
-    // clearedHistory, so the reporter still sends those it has not.
+        if (valid.length === 0) return null;
-    reset() {
+        return Math.round(
-        this.clearedHistory = [...this.clearedHistory, ...this.history].slice(
+            valid.reduce((s, p) => s + p.latency, 0) / valid.length,
            -CONFIG.maxHistoryPoints,
        );
        this.history = [];
        this.lastLatency = null;
        this.status = "pending";
    }
-    // The min, max, median and average latency of the checks in the history
+    minLatency() {
-    // that got an answer.
+        const valid = this.history.filter((p) => p.latency !== null);
-    historyStats() {
+        if (valid.length === 0) return null;
-        return latencyStats(
+        return Math.min(...valid.map((p) => p.latency));
-            this.history
+    }
    maxLatency() {
        const valid = this.history.filter((p) => p.latency !== null);
        if (valid.length === 0) return null;
        return Math.max(...valid.map((p) => p.latency));
    }
    medianLatency() {
        const sorted = this.history
            .filter((p) => p.latency !== null)
-                .map((p) => p.latency),
+            .map((p) => p.latency)
-        );
+            .sort((a, b) => a - b);
        if (sorted.length === 0) return null;
        const mid = Math.floor(sorted.length / 2);
        return sorted.length % 2
            ? sorted[mid]
            : Math.round((sorted[mid - 1] + sorted[mid]) / 2);
    }
    _trim() {
@@ -353,13 +288,33 @@ export class AppState {
    /** WAN-only stats from latest sample (excludes local) */
    wanStats() {
-        const latencies = this.wan
+        const reachable = this.wan.filter((h) => h.lastLatency !== null);
-            .filter((h) => h.lastLatency !== null)
+        const latencies = reachable.map((h) => h.lastLatency);
-            .map((h) => h.lastLatency);
+        const total = this.wan.length;
        if (latencies.length === 0)
            return {
                reachable: 0,
                total,
                min: null,
                max: null,
                med: null,
                avg: null,
            };
        const sorted = [...latencies].sort((a, b) => a - b);
        const mid = Math.floor(sorted.length / 2);
        const med =
            sorted.length % 2
                ? sorted[mid]
                : Math.round((sorted[mid - 1] + sorted[mid]) / 2);
        return {
            reachable: latencies.length,
-            total: this.wan.length,
+            total,
-            ...latencyStats(latencies),
+            min: Math.min(...latencies),
            max: Math.max(...latencies),
            med,
            avg: Math.round(
                latencies.reduce((a, b) => a + b, 0) / latencies.length,
            ),
        };
    }
@@ -385,16 +340,12 @@ export class AppState {
        const timeouts = this.wan.filter(
            (h) => h.status === "error" || h.status === "offline",
        ).length;
-        if (
+        if (timeouts > 10 && reachable <= 4) return "offline";
-            timeouts > CONFIG.offlineTimeouts &&
+        if (timeouts > 4) return "degraded";
            reachable <= CONFIG.offlineReachable
        )
            return "offline";
        if (timeouts > CONFIG.degradedTimeouts) return "degraded";
        const slow = this.wan.filter(
-            (h) => h.lastLatency !== null && h.lastLatency > CONFIG.slowLatency,
+            (h) => h.lastLatency !== null && h.lastLatency > 1000,
        ).length;
-        if (slow > CONFIG.slowHosts) return "slow";
+        if (slow > 3) return "slow";
        return "healthy";
    }
@@ -453,12 +404,9 @@ function getClientId() {
 // Build the delta report body the backend decodes, plus the new per-host
 // high-water marks. Pure function of the passed state: `hosts` is an array
-// of { name, url, status, clearedHistory, history }, `since` maps a host url
+// of { name, url, status, history }, `since` maps a host url to the Unix-ms
-// to the Unix-ms timestamp of the last sample already reported for it, and
+// timestamp of the last sample already reported for it, and `now` is a Date.
-// `now` is a Date. Only non-paused samples newer than the mark are included,
+// Only non-paused samples newer than the mark are included. Returns null
 // from clearedHistory as well as history. A host whose status is pending,
 // as the interval menu leaves it until its next check, is sent with the
 // status of its last check, which is the last sample sent. Returns null
 // when no host has an unreported sample.
 export function buildReport(hosts, clientId, now, since) {
    const reportHosts = [];
@@ -467,7 +415,7 @@ export function buildReport(hosts, clientId, now, since) {
        const mark = since.get(host.url) ?? 0;
        const samples = [];
        let high = mark;
-        for (const p of [...host.clearedHistory, ...host.history]) {
+        for (const p of host.history) {
            if (p.paused) continue;
            if (p.timestamp <= mark) continue;
            samples.push({
@@ -481,10 +429,7 @@ export function buildReport(hosts, clientId, now, since) {
        reportHosts.push({
            name: host.name,
            url: host.url,
-            status:
+            status: host.status,
                host.status === "pending"
                    ? statusFromError(samples.at(-1).error)
                    : host.status,
            history: samples,
        });
        marks.set(host.url, high);
@@ -511,7 +456,7 @@ export function buildReport(hosts, clientId, now, since) {
 // Only one report is ever in flight, and it is abandoned after half the
 // interval, so a slow POST can neither overlap the next report (which would
 // carry the same samples) nor stall reporting for good.
-export class Reporter {
+class Reporter {
    constructor(state, clientId, intervalMs) {
        this.state = state;
        this.clientId = clientId;
@@ -569,13 +514,7 @@ export class Reporter {
 // Checks one target. The check times out after CONFIG.requestTimeout; the
 // caller can give it up sooner through the optional signal, which also ends
-// it as a timeout. A failed check's reason says what went wrong and after
+// it as a timeout.
 // how long; a check that answered has none.
 //
 // The URL is fetched as written, with nothing added to it: the Hetzner
 // speed-test servers close the connection without an answer when the URL
 // has a query string, and cache: "no-store" keeps the browser's cache out
 // of the measurement.
 export async function measureLatency(url, signal) {
    const controller = new AbortController();
    const timeoutId = setTimeout(
@@ -584,10 +523,13 @@ export async function measureLatency(url, signal) {
    );
    signal?.addEventListener("abort", () => controller.abort());
    const targetUrl = new URL(url);
    targetUrl.searchParams.set("_cb", Date.now().toString());
    const start = performance.now();
    try {
-        await fetch(url, {
+        await fetch(targetUrl.toString(), {
            method: "GET",
            mode: "no-cors",
            cache: "no-store",
@@ -596,28 +538,18 @@ export async function measureLatency(url, signal) {
        const latency = Math.round(performance.now() - start);
        clearTimeout(timeoutId);
        if (latency > CONFIG.maxLatency) {
-            return {
+            log.error(`${url} timeout (${latency}ms > ${CONFIG.maxLatency}ms)`);
-                latency: null,
+            return { latency: null, error: "timeout" };
                error: "timeout",
                reason: `answered after ${latency} ms, over the ${CONFIG.maxLatency} ms limit`,
            };
        }
-        return { latency, error: null, reason: null };
+        return { latency, error: null };
    } catch (err) {
        const took = Math.round(performance.now() - start);
        clearTimeout(timeoutId);
        if (err.name === "AbortError") {
-            return {
+            log.error(`${url} timeout (aborted)`);
-                latency: null,
+            return { latency: null, error: "timeout" };
                error: "timeout",
                reason: `timed out after ${took} ms (limit ${CONFIG.requestTimeout} ms)`,
            };
        }
-        return {
+        log.error(`${url} unreachable`);
-            latency: null,
+        return { latency: null, error: "unreachable" };
            error: "unreachable",
            reason: `network error (${err.name}: ${err.message}) after ${took} ms`,
        };
    }
 }
@@ -625,13 +557,21 @@ export async function measureLatency(url, signal) {
 export function latencyHex(latency) {
    if (latency === null) return "#6b7280";
-    return CONFIG.latencyColors.find((c) => latency < c.below).hex;
+    if (latency < 50) return "#22c55e";
    if (latency < 100) return "#84cc16";
    if (latency < 200) return "#eab308";
    if (latency < 500) return "#f97316";
    return "#ef4444";
 }
 export function latencyClass(latency, status) {
    if (status === "offline" || status === "error" || latency === null)
        return "text-gray-500";
-    return CONFIG.latencyColors.find((c) => latency < c.below).className;
+    if (latency < 50) return "text-green-500";
    if (latency < 100) return "text-lime-500";
    if (latency < 200) return "text-yellow-500";
    if (latency < 500) return "text-orange-500";
    return "text-red-500";
 }
 // --- Sparkline Renderer ------------------------------------------------------
@@ -905,7 +845,7 @@ function buildUI(state) {
      </div>
      <footer class="mt-8 text-center text-gray-600 text-xs">
-        <p>Latency measured via GET requests | CORS restrictions may affect some measurements</p>
+        <p>Latency measured via GET requests | IPv4 only | CORS restrictions may affect some measurements</p>
        <p class="mt-2">
          <span class="inline-block w-3 h-3 rounded-full bg-green-500 mr-1 align-middle"></span>&lt;50ms
          <span class="inline-block w-3 h-3 rounded-full bg-lime-500 mr-1 ml-3 align-middle"></span>&lt;100ms
@@ -972,7 +912,10 @@ function updateHostRow(host, index) {
        latencyEl.innerHTML = `<span class="text-gray-500">---</span>`;
    }
-    const { min, med, avg, max } = host.historyStats();
+    const avg = host.averageLatency();
    const med = host.medianLatency();
    const min = host.minLatency();
    const max = host.maxLatency();
    if (host.status === "online" && avg !== null) {
        statusEl.innerHTML = statusStatsHTML([
            ["min", min],
@@ -1109,8 +1052,9 @@ function updateHealthBox(state) {
 // --- Debug Log Renderer ------------------------------------------------------
-// One line of the debug log panel, its message set as text.
+function renderDebugLog() {
-function debugLogLine(entry) {
+    const el = document.getElementById("debug-log");
    if (!el) return;
    const levelColors = {
        error: "text-red-400",
        warning: "text-orange-400",
@@ -1118,43 +1062,19 @@ function debugLogLine(entry) {
        info: "text-gray-300",
        debug: "text-gray-500",
    };
    el.replaceChildren(
        ...debugLog.map((entry) => {
            const ts = formatUTCTimestamp(entry.timestamp);
            const lvl = entry.level.toUpperCase().padEnd(7);
            const line = document.createElement("div");
            line.className = levelColors[entry.level] || "text-gray-400";
            line.textContent = `${ts} ${lvl} ${entry.message}`;
            return line;
-}
+        }),
-
+    );
 // Draws every line of the debug log into the panel, when it is opened.
 function renderDebugLog() {
    const el = document.getElementById("debug-log");
    if (!el) return;
    el.replaceChildren(...debugLog.map((entry) => debugLogLine(entry)));
    el.scrollTop = el.scrollHeight;
 }
 // The request that scrolls the open panel to its newest line in the next
 // frame. One waits at most, also in a hidden tab, where frames do not run.
 let debugLogScrollFrame = 0;
 // Adds entry to the open panel as its last line, and drops the first line
 // once the panel holds more than maxLogEntries. A check that ends while the
 // page is busy is timed only when the page is done, so the lines already
 // shown are left as they are, and the panel is scrolled once a frame
 // rather than once a line: scrolling it lays out all of its lines again.
 function appendDebugLogLine(entry) {
    const el = document.getElementById("debug-log");
    if (!el) return;
    el.append(debugLogLine(entry));
    if (el.children.length > CONFIG.maxLogEntries)
        el.removeChild(el.firstElementChild);
    cancelAnimationFrame(debugLogScrollFrame);
    debugLogScrollFrame = requestAnimationFrame(() => {
        el.scrollTop = el.scrollHeight;
    });
 }
 // --- Sorting -----------------------------------------------------------------
 // Sort WAN hosts: pinned hosts first (alphabetically), then unpinned hosts
@@ -1207,26 +1127,6 @@ function sortAndRebuildWAN(state) {
 // --- Main Loop ---------------------------------------------------------------
 // Writes one line to the browser console, and the same line to the debug
 // log, for a check the page is about to record: for every check that
 // failed, and for a check that answered after checks that failed. Called
 // before host.pushSample, while host.consecutiveFailures still counts the
 // checks before this one.
 function logCheck(host, result) {
    const target = `${host.name} ${host.url} at ${new Date().toISOString()}`;
    const failed = host.consecutiveFailures;
    if (result.error) {
        const line = `netwatch: check failed: ${target}: ${result.reason}`;
        console.error(line);
        log.error(line);
    } else if (failed > 0) {
        const checks = failed === 1 ? "check" : "checks";
        const line = `netwatch: target recovered: ${target}: answered after ${result.latency} ms, following ${failed} failed ${checks} in a row`;
        console.info(line);
        log.info(line);
    }
 }
 export async function tick(state, signal, onOffline) {
    const ts = Date.now();
@@ -1260,7 +1160,6 @@ export async function tick(state, signal, onOffline) {
            if (state.paused || signal.aborted || state.tickCount === 0) {
                return;
            }
            logCheck(host, r);
            host.pushSample(ts, r);
            updateHostRow(host, state.allHosts.indexOf(host));
            log.debug(`${host.name}: ${r.error ? r.error : r.latency + "ms"}`);
@@ -1283,9 +1182,8 @@ export async function tick(state, signal, onOffline) {
    // rows whose check ended before the resume still read "paused"
    state.allHosts.forEach((host, i) => updateHostRow(host, i));
-    // Sort after the first real check, then every CONFIG.roundsPerSort
+    // Sort after the first real check, then every 10 ticks thereafter
-    // ticks thereafter
+    if (state.tickCount === 2 || state.tickCount % 10 === 1) {
    if (state.tickCount === 2 || state.tickCount % CONFIG.roundsPerSort === 1) {
        sortAndRebuildWAN(state);
    }
@@ -1308,10 +1206,9 @@ export async function tick(state, signal, onOffline) {
 // --- Recovery Probe ----------------------------------------------------------
-// When offline, check CONFIG.recoveryProbeHosts random WAN hosts every
+// When offline, check 4 random WAN hosts every 500ms, giving up the checks
-// CONFIG.recoveryProbeInterval ms, giving up the checks started one interval
+// started 500ms before, so at most 4 are ever waiting. As soon as one
-// before, so at most that many are ever waiting. As soon as one answers,
+// answers, stop probing and start a new round at once.
 // stop probing and start a new round at once.
 function startRecoveryProbe(state, startRounds) {
    if (state._recoveryProbeId) return; // already running
    const candidates = [...state.wan];
@@ -1319,7 +1216,7 @@ function startRecoveryProbe(state, startRounds) {
        const j = Math.floor(Math.random() * (i + 1));
        [candidates[i], candidates[j]] = [candidates[j], candidates[i]];
    }
-    const canaries = candidates.slice(0, CONFIG.recoveryProbeHosts);
+    const canaries = candidates.slice(0, 4);
    log.notice(
        `Recovery probe started (${canaries.map((h) => h.name).join(", ")})`,
    );
@@ -1336,7 +1233,7 @@ function startRecoveryProbe(state, startRounds) {
                startRounds();
            });
        }
-    }, CONFIG.recoveryProbeInterval);
+    }, 500);
 }
 function stopRecoveryProbe(state) {
@@ -1555,7 +1452,9 @@ async function init() {
            // Clear all host history and reset state
            for (const host of state.allHosts) {
-                host.reset();
+                host.history = [];
                host.lastLatency = null;
                host.status = "pending";
            }
            state.tickCount = 0;
@@ -1598,7 +1497,7 @@ async function init() {
        });
    window.addEventListener("resize", () => handleResize(state));
-    setTimeout(() => handleResize(state), CONFIG.resizeDelay);
+    setTimeout(() => handleResize(state), 100);
 }
 // Bootstrap only when loaded as the page: a real DOM containing the #app
@@ -1,4 +1,4 @@
-// Unit tests for web/src/main.js, run with Node's built-in test runner by the
+// Unit tests for src/main.js, run with Node's built-in test runner by the
 // test script in package.json. Importing the module does not start the
 // page.
@@ -13,29 +13,22 @@ import {
    humanDuration,
    latencyClass,
    latencyHex,
    log,
    measureLatency,
    Reporter,
    tick,
-} from "../../web/src/main.js";
+} from "../../src/main.js";
 // There is no page here, so the tests stand in for it. The debug log looks
-// for its panel by id and finds none, an element the page creates is a
+// for its panel by id and finds none. Each element of a host's row that
-// plain object, and no frame the page asks for ever comes. Each element of
+// tick or greyOutUI draws into is a plain object, made the first time a
-// a host's row that tick or greyOutUI draws into is a plain object, made
+// test looks it up and kept in elements under its selector until the next
-// the first time a test looks it up and kept in elements under its
+// test starts. As on a page, writing its text replaces its markup; the
-// selector until the next test starts. As on a page, writing its text
+// status dot greyOutUI looks for in it is not there. Drawing a sparkline
-// replaces its markup; the status dot greyOutUI looks for in it is not
+// does nothing; it looks for the pixel ratio on window and finds none.
 // there. Drawing a sparkline does nothing; it looks for the pixel ratio on
 // window and finds none. In each test, console.error and console.info
 // print nothing and keep what they are given.
 const doNothing = () => {};
 let elements;
-beforeEach((t) => {
+beforeEach(() => {
    elements = {};
    t.mock.method(console, "error", doNothing);
    t.mock.method(console, "info", doNothing);
 });
 const doNothing = () => {};
 const canvasContext = {
    clearRect: doNothing,
    beginPath: doNothing,
@@ -48,11 +41,8 @@ const canvasContext = {
    arc: doNothing,
 };
 globalThis.window = {};
 globalThis.requestAnimationFrame = doNothing;
 globalThis.cancelAnimationFrame = doNothing;
 globalThis.document = {
    getElementById: () => null,
    createElement: () => ({}),
    querySelector: (selector) =>
        (elements[selector] ??= {
            getContext: () => canvasContext,
@@ -79,10 +69,8 @@ function statusText(state, host) {
 // Mocks the clock for test t, so that a check lasting seconds takes no real
 // time, and replaces fetch with targets that each answer after
 // answerAfter(url) milliseconds of that clock, or never when that is
-// Infinity. The target at unreachableUrl, if one is given, does not
+// Infinity. Both are restored when the test ends.
-// answer: after that time its fetch fails with the error a browser gives
+function mockTargets(t, answerAfter) {
 // the page for a network error. Both are restored when the test ends.
 function mockTargets(t, answerAfter, unreachableUrl) {
    t.mock.timers.enable({ apis: ["setTimeout", "Date"] });
    t.mock.method(performance, "now", () => Date.now());
    t.mock.method(
@@ -90,12 +78,8 @@ function mockTargets(t, answerAfter, unreachableUrl) {
        "fetch",
        (url, { signal }) =>
            new Promise((resolve, reject) => {
                const answer =
                    url === unreachableUrl
                        ? () => reject(new TypeError("Failed to fetch"))
                        : resolve;
                if (answerAfter(url) !== Infinity) {
-                    setTimeout(answer, answerAfter(url));
+                    setTimeout(resolve, answerAfter(url));
                }
                signal.addEventListener("abort", () => reject(signal.reason));
            }),
@@ -123,7 +107,6 @@ for (const interval of [10000, 30000]) {
        assert.deepEqual(await settled(check), {
            latency: slowAnswer,
            error: null,
            reason: null,
        });
    });
@@ -137,37 +120,10 @@ for (const interval of [10000, 30000]) {
        assert.deepEqual(await settled(check), {
            latency: null,
            error: "timeout",
            reason: `timed out after ${timeout} ms (limit ${timeout} ms)`,
        });
    });
 }
 // A browser can run a timer late, on a busy page or in a background tab.
 // Moving the mocked clock on 30000ms at once runs the 24000ms timeout with
 // the clock already at 30000ms.
 test("at a 30000ms interval, a check whose timeout runs late gives how long the request took and the time limit", async (t) => {
    CONFIG.updateInterval = 30000;
    mockTargets(t, () => Infinity);
    const check = measureLatency("https://target.test");
    t.mock.timers.tick(30000);
    assert.deepEqual(await settled(check), {
        latency: null,
        error: "timeout",
        reason: "timed out after 30000 ms (limit 24000 ms)",
    });
 });
 test("a check fetches the target's URL as written, with no query string added", async (t) => {
    mockTargets(t, () => 10);
    const check = measureLatency("https://fsn1-speed.hetzner.com");
    t.mock.timers.tick(10);
    assert.notEqual(await settled(check), "still waiting");
    assert.equal(
        fetch.mock.calls[0].arguments[0],
        "https://fsn1-speed.hetzner.com",
    );
 });
 test("at a 30000ms interval, a target answering after 1000ms shows in its row while another target's check is still waiting", async (t) => {
    CONFIG.updateInterval = 30000;
    const state = new AppState([
@@ -176,7 +132,7 @@ test("at a 30000ms interval, a target answering after 1000ms shows in its row wh
    const answering = state.local[0];
    const waiting = state.wan[0];
    // No target but the answering one ever answers.
-    mockTargets(t, (url) => (url === answering.url ? 1000 : Infinity));
+    mockTargets(t, (url) => (url.startsWith(answering.url) ? 1000 : Infinity));
    // The third tick: the first is discarded as a whole, and the second ends
    // by sorting the rows, which rebuilds a page that is not here.
    state.tickCount = 2;
@@ -203,7 +159,7 @@ test("at a 30000ms interval, a check still waiting when its round is given up do
        { name: "Answering", url: "https://answering.test" },
    ]);
    const answering = state.local[0];
-    mockTargets(t, (url) => (url === answering.url ? 1000 : Infinity));
+    mockTargets(t, (url) => (url.startsWith(answering.url) ? 1000 : Infinity));
    state.tickCount = 2;
    const roundChecks = new AbortController();
@@ -221,7 +177,7 @@ test("at a 30000ms interval, a check still waiting when the user pauses does not
        { name: "Answering", url: "https://answering.test" },
    ]);
    const answering = state.local[0];
-    mockTargets(t, (url) => (url === answering.url ? 1000 : Infinity));
+    mockTargets(t, (url) => (url.startsWith(answering.url) ? 1000 : Infinity));
    state.tickCount = 2;
    const round = tick(state, new AbortController().signal);
@@ -238,7 +194,7 @@ test("at a 30000ms interval, a check in the first round does not show in its row
        { name: "Answering", url: "https://answering.test" },
    ]);
    const answering = state.local[0];
-    mockTargets(t, (url) => (url === answering.url ? 1000 : Infinity));
+    mockTargets(t, (url) => (url.startsWith(answering.url) ? 1000 : Infinity));
    const round = tick(state, new AbortController().signal);
    t.mock.timers.tick(1000);
@@ -252,7 +208,7 @@ test("at a 30000ms interval, after the user pauses and resumes during a round, n
        { name: "Answering", url: "https://answering.test" },
    ]);
    const answering = state.local[0];
-    mockTargets(t, (url) => (url === answering.url ? 1000 : Infinity));
+    mockTargets(t, (url) => (url.startsWith(answering.url) ? 1000 : Infinity));
    state.tickCount = 2;
    const round = tick(state, new AbortController().signal);
@@ -275,240 +231,6 @@ test("at a 30000ms interval, after the user pauses and resumes during a round, n
    }
 });
 // In the next tests a round checks one target, Target, at a 30000ms
 // interval, so a check times out after 24000ms. The mocked clock starts at
 // 1970-01-01T00:00:00.000Z.
 // An app state with Target and no WAN targets, so its rounds check only
 // Target, and whose next round is recorded: it is the third, as the first
 // is discarded and the second ends by sorting the rows, which rebuilds a
 // page that is not here. Returns it and Target.
 function stateWithOneTarget() {
    CONFIG.updateInterval = 30000;
    const state = new AppState([
        { name: "Target", url: "https://target.test" },
    ]);
    state.wan = [];
    state.tickCount = 2;
    return { state, target: state.local[0] };
 }
 // What this test wrote to the browser console with console[method].
 function consoleLines(method) {
    return console[method].mock.calls.map((call) => call.arguments[0]);
 }
 test("a check that fails with a network error writes one console line with the target, the time, the error and how long the request took", async (t) => {
    const { state, target } = stateWithOneTarget();
    mockTargets(t, () => 23, target.url);
    const round = tick(state, new AbortController().signal);
    t.mock.timers.tick(23);
    assert.notEqual(await settled(round), "still waiting");
    assert.deepEqual(consoleLines("error"), [
        "netwatch: check failed: Target https://target.test at 1970-01-01T00:00:00.023Z: network error (TypeError: Failed to fetch) after 23 ms",
    ]);
    assert.deepEqual(consoleLines("info"), []);
 });
 test("a check that times out writes one console line with the target, the time, how long the request took and the time limit", async (t) => {
    const { state } = stateWithOneTarget();
    mockTargets(t, () => Infinity);
    const round = tick(state, new AbortController().signal);
    t.mock.timers.tick(24000);
    assert.notEqual(await settled(round), "still waiting");
    assert.deepEqual(consoleLines("error"), [
        "netwatch: check failed: Target https://target.test at 1970-01-01T00:00:24.000Z: timed out after 24000 ms (limit 24000 ms)",
    ]);
    assert.deepEqual(consoleLines("info"), []);
 });
 // An answer that took longer than CONFIG.maxLatency is recorded as a
 // timeout. The limit is the check's own timeout, so such an answer is one
 // that came in just as the check timed out; here it is lowered to 500ms.
 test("a check answered over the time limit writes one console line with the target, the time, how long the answer took and the limit", async (t) => {
    const { state } = stateWithOneTarget();
    t.mock.getter(CONFIG, "maxLatency", () => 500);
    mockTargets(t, () => 1000);
    const round = tick(state, new AbortController().signal);
    t.mock.timers.tick(1000);
    assert.notEqual(await settled(round), "still waiting");
    assert.deepEqual(consoleLines("error"), [
        "netwatch: check failed: Target https://target.test at 1970-01-01T00:00:01.000Z: answered after 1000 ms, over the 500 ms limit",
    ]);
    assert.deepEqual(consoleLines("info"), []);
 });
 test("a check that answers writes nothing to the console", async (t) => {
    const { state } = stateWithOneTarget();
    mockTargets(t, () => 30);
    const round = tick(state, new AbortController().signal);
    t.mock.timers.tick(30);
    assert.notEqual(await settled(round), "still waiting");
    assert.deepEqual(consoleLines("error"), []);
    assert.deepEqual(consoleLines("info"), []);
 });
 for (const [failed, checks] of [
    [1, "1 failed check"],
    [3, "3 failed checks"],
 ]) {
    test(`a check that answers after ${checks} writes one console line saying the target recovered, and the next writes nothing`, async (t) => {
        const { state, target } = stateWithOneTarget();
        for (let i = 0; i < failed; i++) {
            target.pushSample(Date.now(), {
                latency: null,
                error: "unreachable",
            });
        }
        mockTargets(t, () => 40);
        for (let i = 0; i < 2; i++) {
            const round = tick(state, new AbortController().signal);
            t.mock.timers.tick(40);
            assert.notEqual(await settled(round), "still waiting");
        }
        assert.deepEqual(consoleLines("info"), [
            `netwatch: target recovered: Target https://target.test at 1970-01-01T00:00:00.040Z: answered after 40 ms, following ${checks} in a row`,
        ]);
        assert.deepEqual(consoleLines("error"), []);
    });
 }
 test("a check that fails in the first round, which is discarded, writes nothing to the console", async (t) => {
    const { state, target } = stateWithOneTarget();
    state.tickCount = 0;
    mockTargets(t, () => 23, target.url);
    const round = tick(state, new AbortController().signal);
    t.mock.timers.tick(23);
    assert.notEqual(await settled(round), "still waiting");
    assert.deepEqual(consoleLines("error"), []);
    assert.deepEqual(consoleLines("info"), []);
 });
 // The reports Reporter sent in this test, each as the hosts it sent.
 function reportedHosts() {
    return fetch.mock.calls.map(
        (call) => JSON.parse(call.arguments[1].body).hosts,
    );
 }
 // In the next two tests every report is delivered, and only Target has
 // checks to report. target.reset() does to Target what the interval menu
 // does to each target.
 // The interval changes after the first report. The second report is sent
 // before Target's first check since then, and the third after it. A fourth
 // would hold no check the first three did not, so it is not sent.
 test("checks recorded before an interval change and not yet reported go in the next report, with the status of the last check, and no check is reported twice", async (t) => {
    const state = new AppState([
        { name: "Target", url: "https://target.test" },
    ]);
    const target = state.local[0];
    const reporter = new Reporter(state, "client", CONFIG.reportInterval);
    t.mock.method(globalThis, "fetch", async () => ({ ok: true }));
    target.pushSample(1000, { latency: 10, error: null });
    await reporter.flush();
    target.pushSample(2000, { latency: 20, error: null });
    target.pushSample(3000, { latency: null, error: "timeout" });
    target.reset();
    await reporter.flush();
    target.pushSample(4000, { latency: 40, error: null });
    await reporter.flush();
    await reporter.flush();
    const name = "Target";
    const url = "https://target.test";
    assert.deepEqual(reportedHosts(), [
        [
            {
                name,
                url,
                status: "online",
                history: [{ t: 1000, latency: 10, error: null }],
            },
        ],
        [
            {
                name,
                url,
                status: "error",
                history: [
                    { t: 2000, latency: 20, error: null },
                    { t: 3000, latency: null, error: "timeout" },
                ],
            },
        ],
        [
            {
                name,
                url,
                status: "online",
                history: [{ t: 4000, latency: 40, error: null }],
            },
        ],
    ]);
    // The page's history starts over.
    assert.deepEqual(target.history, [
        { timestamp: 4000, latency: 40, error: null },
    ]);
 });
 // The interval changes twice, after 60 checks each time, before the first
 // report. Of the 120 checks the interval menu cleared, the report has the
 // last 100 (CONFIG.maxHistoryPoints).
 test("of the checks interval changes cleared before they were reported, the next report has the last 100", async (t) => {
    const state = new AppState([
        { name: "Target", url: "https://target.test" },
    ]);
    const target = state.local[0];
    const reporter = new Reporter(state, "client", CONFIG.reportInterval);
    t.mock.method(globalThis, "fetch", async () => ({ ok: true }));
    for (let n = 1; n <= 120; n++) {
        target.pushSample(n * 1000, { latency: 10, error: null });
        if (n % 60 === 0) target.reset();
    }
    await reporter.flush();
    const sent = reportedHosts()[0][0].history;
    assert.equal(sent.length, 100);
    assert.equal(sent[0].t, 21000);
    assert.equal(sent.at(-1).t, 120000);
 });
 // panel stands in for the open debug log panel, and for the element in it
 // that holds its lines, whichever the page looks up by id; the lines are
 // kept in order in lines. Drawing the whole panel, as opening it does,
 // replaces every line with a new one.
 test("with the debug log open, writing more lines than it keeps adds each to the panel and leaves the lines already shown as they are", (t) => {
    const lines = [];
    const panel = {
        classList: { contains: () => false },
        children: lines,
        get firstElementChild() {
            return lines[0];
        },
        append: (line) => lines.push(line),
        removeChild: (line) => lines.splice(lines.indexOf(line), 1),
        replaceChildren: (...newLines) =>
            lines.splice(0, lines.length, ...newLines),
    };
    t.mock.method(globalThis.document, "getElementById", () => panel);
    const kept = CONFIG.maxLogEntries;
    for (let i = 1; i <= kept; i++) log.info(`line ${i}`);
    const shown = [...lines];
    for (let i = kept + 1; i <= kept + 5; i++) log.info(`line ${i}`);
    assert.deepEqual(
        lines.map((line) => line.textContent.match(/line \d+$/)?.[0]),
        Array.from({ length: kept }, (_, i) => `line ${i + 6}`),
    );
    // The first five lines shown were dropped, and the rest are the same
    // lines, not drawn again.
    shown.slice(5).forEach((line, i) => assert.equal(lines[i], line));
 });
 // The page shows &lt; &gt; &quot; &amp; and &#39; in a row's markup as
 // < > " & and '.
 test(`a target whose name and URL hold < > " & and ' shows those characters in its row`, () => {
@@ -625,35 +347,19 @@ for (const { history, latencies, statistics } of [
    },
 ]) {
    test(`a target's min, max, average and median latency over ${history}`, () => {
-        const { min, max, avg, med } = hostAfter(latencies).historyStats();
+        const host = hostAfter(latencies);
-        assert.deepEqual({ min, max, average: avg, median: med }, statistics);
+        assert.deepEqual(
            {
                min: host.minLatency(),
                max: host.maxLatency(),
                average: host.averageLatency(),
                median: host.medianLatency(),
            },
            statistics,
        );
    });
 }
 // The summary's figures come from each WAN target's last check, by the same
 // rules as a target's own: here four answered, one was found unreachable
 // and the rest have not been checked yet. The median, 22.5, and the
 // average, 21.25, are rounded.
 test("the summary's min, max, median and average latency over the WAN targets' last checks", () => {
    const state = new AppState([]);
    [30, 10, null, 25, 20].forEach((latency, i) =>
        state.wan[i].pushSample(
            Date.now(),
            latency === null
                ? { latency: null, error: "unreachable" }
                : { latency, error: null },
        ),
    );
    assert.deepEqual(state.wanStats(), {
        reachable: 4,
        total: state.wan.length,
        min: 10,
        max: 30,
        med: 23,
        avg: 21,
    });
 });
 // An app state in which, of the WAN targets, the first timedOut timed out,
 // the next unreachable were found unreachable, the next answered answered
 // after latency ms, and the rest have not been checked yet.
@@ -7,22 +7,22 @@ make frontend-viewport-test
 ```
 It builds `dist/`, serves it from the same digest-pinned `nginx` image and the
-same `configs/nginx.conf` the shipping container uses, drives a digest-pinned
+same `nginx.conf` the shipping container uses, drives a digest-pinned headless
-headless Chrome against it over CDP, and asserts on computed layout at every
+Chrome against it over CDP, and asserts on computed layout at every viewport
-viewport width derived from the app's own CSS. Screenshots land in
+width derived from the app's own CSS. Screenshots land in `tmp/viewport/`
-`tmp/viewport/` alongside a `results.json`; they are artifacts for a human to
+alongside a `results.json`; they are artifacts for a human to look at when
-look at when something fails, not the evidence. The assertions are the evidence.
+something fails, not the evidence. The assertions are the evidence.
-The target is deliberately outside `make check`: it takes minutes, and
+The target is deliberately outside `make check`: it needs Docker and takes
-`make test` has to stay under 60 seconds.
+minutes, and `make test` has to stay under 20 seconds.
 ## How the widths are chosen
 Not from a list of phone models. `viewports.js` parses the `@media` conditions
-out of `web/src/styles.css` and scans `web/src/main.js` and `web/index.html` for
+out of `src/styles.css` and scans `src/main.js` and `index.html` for Tailwind
-Tailwind responsive prefixes, then tests every breakpoint it finds at one pixel
+responsive prefixes, then tests every breakpoint it finds at one pixel below it,
-below it, exactly on it, and one pixel above it. A generic 375px "phone" test
+exactly on it, and one pixel above it. A generic 375px "phone" test sails
-sails straight past an off-by-one at a media query boundary; `max-width: 768px`
+straight past an off-by-one at a media query boundary; `max-width: 768px`
 matches _at_ 768, and the sweep pins down which side of that line each layout is
 on.
@@ -77,7 +77,7 @@ the internet, so the app's latency probes cannot reach anything real. The
 harness answers them itself from a fixed delay table, with a deterministic
 fraction failed outright, so the rows render a realistic spread of one-, two-
 and three-digit latencies plus some unreachable rows. That spread is what the
-layout has to survive; a `---` placeholder in every row would not exercise it.
+layout has to survive; 24 identical `---` placeholders would not exercise it.
 ## What this cannot verify
@@ -104,11 +104,10 @@ Everything else this issue was actually about — does the layout reflow, does
 anything overflow, is content clipped, are the controls big enough — is a
 function of viewport width and CSS, and is covered above.
-## Relation to the unit tests
+## Relation to the unit test framework (#21)
-Complementary layers, not two stacks. The unit tests in `test/unit/`, which
+Complementary layers, not two stacks. `vitest` (#21) will exercise module-level
-`make test` runs with Node's built-in test runner, exercise the functions
+logic in-process with no browser. This harness exercises rendered layout in a
-`web/src/main.js` exports in-process with no browser. This harness exercises
+real engine and is the only thing here that can see a media query. Neither
-rendered layout in a real engine and is the only thing here that can see a media
+replaces the other; assertions about computed styles and element geometry belong
-query. Neither replaces the other; assertions about computed styles and element
+here, assertions about functions belong in `vitest`.
 geometry belong here, assertions about functions belong in `test/unit/`.
@@ -51,8 +51,7 @@ const PROBE_DELAYS_MS = [2, 45, 123, 456, 780];
 // offline row rendering is exercised too.
 const UNREACHABLE_MODULUS = 7;
 // The gateway candidate that "answers", so gateway detection succeeds and
-// the Local Gateway row renders. Matches GATEWAY_CANDIDATES in
+// the Local Gateway row renders. Matches GATEWAY_CANDIDATES in src/main.js.
 // web/src/main.js.
 const RESPONSIVE_GATEWAY = "http://192.168.1.1";
 const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
@@ -161,27 +161,26 @@ const ANCHOR_VIEWPORTS = [
 export function deriveViewports(root) {
    const conditions = [
        ...mediaConditionsFromCss(
-            readFileSync(join(root, "web/src/styles.css"), "utf8"),
+            readFileSync(join(root, "src/styles.css"), "utf8"),
-            "web/src/styles.css",
+            "src/styles.css",
        ),
        ...mediaConditionsFromMarkup([
            {
-                path: "web/src/main.js",
+                path: "src/main.js",
-                text: readFileSync(join(root, "web/src/main.js"), "utf8"),
+                text: readFileSync(join(root, "src/main.js"), "utf8"),
            },
            {
-                path: "web/index.html",
+                path: "index.html",
-                text: readFileSync(join(root, "web/index.html"), "utf8"),
+                text: readFileSync(join(root, "index.html"), "utf8"),
            },
        ]),
    ];
    if (conditions.length === 0) {
        throw new Error(
-            "no responsive breakpoints found in web/src/styles.css, " +
+            "no responsive breakpoints found in src/styles.css, src/main.js or " +
-                "web/src/main.js or web/index.html — either the responsive " +
+                "index.html — either the responsive layout was deleted or this " +
-                "layout was deleted or this derivation has stopped matching " +
+                "derivation has stopped matching the sources",
                "the sources",
        );
    }
@@ -6,10 +6,6 @@ const commitHash = execSync("git rev-parse --short HEAD").toString().trim();
 const commitFull = execSync("git rev-parse HEAD").toString().trim();
 export default defineConfig({
    // The page's source is in web/; the build still goes to dist/ at the
    // repo root, which Dockerfile and the viewport test serve. Vite
    // empties only an outDir inside root unless told to.
    root: "web",
    plugins: [tailwindcss()],
    server: {
        // Proxy /api to a locally running netwatch-server so `yarn dev`
@@ -23,8 +19,6 @@ export default defineConfig({
        __COMMIT_FULL__: JSON.stringify(commitFull),
    },
    build: {
        outDir: "../dist",
        emptyOutDir: true,
        target: "esnext",
        minify: "esbuild",
        cssMinify: true,