DeleteDB stopped at the first file it could not remove, and the
webhook delete handler logged one message for every failure. Once
the database file itself is gone, its events are lost, and telling
the operator only that a file was left behind suggests otherwise.
DeleteDB now tries the database file and both sidecars, and its
error wraps ErrEventDBNotRemoved when the database file is still on
disk, or ErrSidecarNotRemoved when only a -wal or -shm sidecar is
left. The handler logs a different message for each.
Model: opus-5-5
fx printed its dependency graph and lifecycle hooks through its own console logger on standard error, so an operator shipping the JSON log to a collector got a second shape on a second stream for every start. The production app now passes fx.WithLogger with a small FxLogger in internal/logger that writes fx's events through the service's logger: graph events at debug, lifecycle at info, failures at error. Its constructor takes the configuration, so DEBUG=true applies before fx replays the events it held back. go.uber.org/fx moves from v1.20.1 to v1.24.0. Tests keep fx.NopLogger. The README says a failure before the logger exists, and the Go runtime's own output, still go to standard error as plain text.
Model: opus-5-5
A request whose RemoteAddr is empty has no peer identity, so the rate limiters' key falls back to the raw empty string and every such request shares one bucket: it fails closed rather than giving each its own. net/http always fills RemoteAddr for a TCP listener, so normal serving never reaches this. The behaviour is unchanged and now deliberate: a test pins the shared key, and a one-sentence comment at the fallback tells the empty case apart from a Unix-socket listener, where every peer legitimately carries the same address.
Model: opus-5-5
Follow-ups from an August review of the retention bounds, each checked against the current tree. A test now pins that a retention value above the keep-forever sentinel is stored as the sentinel. The form's retention parser returns its message directly, so the two error values that were never compared, and the function that mapped them to messages, are gone. The sweep's own keep-forever skip, which duplicated the check in retentionCutoff, is removed; the cutoff is now asked before the webhook's database is opened. The create-form refill test uses HTML-special characters and checks they come back escaped. The README item was already settled; handling for rows made by hand is declined.
Model: opus-5-5
Behind a trusted proxy every log line named only the proxy, so abuse could not be traced from webhooker's own logs although the rate limiters already knew the client. The access log, the rate-limit rejection lines, the CSRF warning and the receiver's request line now carry clientIP next to remoteIP. remoteIP still means the connecting peer; clientIP is the address the rate limiters key on, the forwarded client when the peer is inside TRUSTED_PROXIES, worked out once per request by the same code. The README says the field is only as trustworthy as TRUSTED_PROXIES. The access log's 2,560-byte line ceiling holds with the field charged, and a size case with an oversized X-Forwarded-For pins it.
Model: opus-5-5
Three retention tests made internal/database the slowest test package, mostly by seeding thousands of rows 500 per insert: the SQLite driver finds each parameter's value by scanning the statement's arguments from the first until it reaches that parameter's, so binding grows with the square of the parameter count. They now seed the same rows 50 per insert, about three times faster; no test case or assertion changes. The package drops from 13 to 22s to about 7s. What keeps make test above the 20s target is now mostly the cold -race compile of the tree, which moves with host load. The 90s per-package timeout stays; script/test's header records the new figures.
Model: opus-5-5
Every page's Content-Security-Policy forbids eval, which the standard Alpine.js build needs, so no directive ran in a browser: both add forms on the webhook page showed open, and events in the event log could not be collapsed. The UI now loads Alpine's CSP build (@alpinejs/csp 3.14.9 in 3p/); the policy is unchanged. Each directive names a property or method of a component registered in static/js/app.js (collapsible, targetForm), and each card holds its own x-data. A browser test, built only with the browser tag, loads the webhook page and the event log under the real headers; make test-browser runs it in Docker. New test-only dependency chromedp, which raises golang.org/x/sys to 0.47.0.
Model: opus-5-5
The archive writer's reopen debounce test made two writes that had to land inside the real 2-second window, then slept 2.1 seconds to cross it, so a slow host could turn correct code red. The archive writer now reads the time for its reopen debounce from a clock field, time.Now in production, and the test moves that clock instead of sleeping: two writes at one instant open the file once, and a write one debounce later closes and reopens it once. Removing the debounce check fails the test. This was the last test whose result depended on real elapsed time.
Model: opus-5-5
Every reference from a model to the record it belongs to (a target's or entrypoint's webhook, a webhook's or API key's user, an event's webhook and entrypoint, a delivery's event and target, a delivery result's delivery) is now tagged json:"-", so a preloaded model can be marshalled without the encoder recursing between parent and child. References to child records stay. Tests marshal each of the nine references set, preloaded where it matters, and check the parent's id is absent, so deleting or restoring any one tag fails a test; preloading still fills each reference.
Model: opus-5-5
Six form-error paths (the login error and the webhook form's validation and name-taken branches) called WriteHeader before rendering, so if the form page's own template failed, the answer kept its 400 or 409 status with an error body instead of a 500. The new renderTemplateStatus renders into the buffer and writes the status only after the page has rendered; renderTemplate sends 200 through it, and those handlers pass their status to it. No handler calls WriteHeader before a render. Login-form tests show the 400 page still renders in full and a failing template answers 500.
Model: opus-5-5
The target toggle loaded the target, flipped its active flag and saved the whole row, so an edit of the same target's name or settings saved in between was written back over and lost, although it reported success. The toggle now updates only the active column, so it can no longer undo an edit. A test saves an edit just after the toggle has read the target and shows the edit survives and the state flips. The entrypoint toggle is unchanged, since an entrypoint has no edit form.
Model: opus-5-5
The middleware tests built their cookie stores by setting store.Options by hand, which left the securecookie codecs at the library's 30-day default instead of the 7-day cap production sets, an invisible divergence that would outlive the next change to store construction. The test store constructor moves from internal/session/export_test.go into internal/session/testing.go so other packages can reach it, and the two middleware test helpers build their stores through it. No test in the repo builds a cookie store by hand any more, and no assertion changes.
Model: opus-5-5
The internal/handlers tests were load-fragile because every test app hashed the admin password at 64 MB; that went with the cheaper test hashing already on next, and measuring under the host's real load found nothing left to fix in how the tests run. The comment on newTestApp now says its start limit, fx's default, is there to catch a start that hangs, and that the slowest measured start is far inside it. The header of script/test gives current figures in place of ones from before that change. Neither limit changes, and no test changes.
Model: opus-5-5
Each entry in the webhook list at /hooks now shows when its last event arrived (or "No events yet"), how many of its deliveries failed in the last 24 hours, in red when not zero, and how many entrypoints and targets are inactive, as in "4 targets, 1 inactive". The figures come from the event totals row and the statistics pane's own query, so the list no longer counts every stored event and its event count matches the pane's. A webhook whose event database cannot be read says so in its entry; the rest of the list still shows. A failed read of entrypoints or targets from the main database now returns an error page instead of showing zero.
Model: opus-5-5
The receiver route /h/{uuid} has no MaxBodySize middleware, unlike the page route groups; its 1 MB cap is enforced in the handler, which owns the response senders see. Nothing at the route said so, so a reader could take the receiver for uncapped or remove the only bound on the one unauthenticated endpoint. The route registration and the handler's body-reading function now say where the cap lives and why. A routing test sends a body of exactly 1 MB and one a byte over through the production router and pins that only the second is refused, with the handler's 413 and message.
Model: opus-5-5
Each database target now writes its own archive file, archive-WEBHOOKNAME-TARGETNAME-TARGETID.db, named by delivery.ArchiveFileName, in place of one archive per webhook keyed on its UUID. Renaming a webhook or a target renames its archive files (with any -wal and -shm) before the new name is saved, never over an existing file, and moves every one back if a rename or the save fails. The webhook edit, the target edit and target creation share one lock so no two interleave. Deleting a webhook or target leaves its files on disk. Nothing looks for the old archive-WEBHOOKID.db files. The README gives the naming and the recovery steps.
Model: opus-5-5
MAINTENANCE_MODE did nothing but make the healthcheck JSON report maintenanceMode: true; no request was ever served differently, so an operator who set it expecting requests to be refused got nothing. It is removed from the configuration, the startup configuration log line, the healthcheck JSON, the README and the Settings page, together with the uncalled Server.MaintenanceMode method and the healthcheck's dependency on the configuration. A leftover value in an environment is ignored like any other unknown variable.
Model: opus-5-5
The access log's response writer and the metrics middleware's writer hid the writer beneath them, so a handler's flush, hijack or write deadline set through http.ResponseController failed with "not supported" behind them. The access log's writer now has Unwrap. The metrics middleware calls the library's public Measure with a writer of our own that has Unwrap, in place of std.Handler's writer, so the middleware order and what metrics record are unchanged. A test over a real connection sets a write deadline and flushes, metrics on and off, on a global route and in an admin page group, and fails without either Unwrap.
Model: opus-5-5
The delivery duration histogram was registered only when a delivery first ran, so until then /metrics had no series for it and a dashboard or alert on it saw nothing at all rather than zero. Each of the four target types now has its series registered at boot with zero counts, so the histogram is served from the first scrape. Tests pin that all four series are present before any delivery, on the metrics registry and through the production router.
Model: opus-5-5
A new page at /settings, linked from the navigation bar and behind the login, lists every configuration field the server loaded at startup: its environment variable, the README table's description, and the value in effect. METRICS_PASSWORD and SENTRY_DSN show only as set or not set; their values are replaced before rendering and never reach the template. The route is GET only and its group is built like the other admin page groups. Tests set the credentials one at a time so each value shown is checked against its own field and a set secret never appears in the page.
Model: opus-5-5
On the build host a connection to [::] reaches a listener on ::1, and one to 0.0.0.0 reaches 127.0.0.1, so a delivery target at either reached this host's loopback past the guard. 0.0.0.0/32 and ::/128 are now in alwaysBlockedNetworks, which no allowlist opens; an allowlist reaches loopback only through an entry covering a loopback address. IPv6 multicast (ff00::/8) and documentation space (2001:db8::/32) are refused by default and reopen when listed.
Every default blocklist entry has a one-line comment, each list is pinned on its own, and tests refuse each address at target creation and at delivery. The README and the rules above each list match.
Model: opus-5-5
Every test that builds an fx app with fxtest.New (handlers, server, resetpw, gormlog, config) now passes fx.NopLogger, so fx's own log no longer goes to t.Logf. A hook still running after a start or stop timeout can then no longer write to a test that has already returned, which the race detector reported as a data race. What the tests assert is unchanged, and nothing about the race detector is suppressed.
Model: opus-5-5
Saving, deleting, activating or deactivating a webhook, entrypoint or target, and signing out, now land on their page with a one-line notice such as "Webhook deleted." or "Signed out.". The redirect carries a fixed code that maps to fixed text; an unknown code shows nothing, so nothing from the URL is ever echoed. One partial in the page layout shows the notice on every page, and replay and resubmit now use the same codes and partial. Error pages show no notice.
Model: opus-5-5
The webhook page's maximum width goes from 72rem (1152 px) to 108rem (1728 px), half again as wide, so an entrypoint URL stays on one line in 1920- and 1440-pixel windows; the statistics pane and both columns widen with it. The title row now wraps, so a phone-width window no longer scrolls sideways.
The width is an inline style: static/css/style.css is linked by no page, and the committed Tailwind stylesheet has no class that wide. The webhook list, the event log, the navbar and the footer stay at 72rem.
Model: opus-5-5
The http and slack targets sent the constant User-Agent webhooker/1.0. They now send webhooker/ followed by the version the build stamped, the same value the footer shows, built in one place on the delivery engine. User-Agent stays a reserved header and is still set after the target's configured headers, so a configured one cannot override it. Tests check the header each target sends against a known version, and that a configured User-Agent is replaced.
Model: opus-5-5
A second metrics-enabled router in one process panicked on a duplicate collector registration, because every collector registered on Prometheus's global default registry. metrics.NewRegistry now builds one registry with the Go runtime and process collectors; fx provides it and the delivery metric set built on it. The middleware builds its HTTP recorder once on that registry (NewForTest on a fresh one), the engine and handlers take the metric set from fx, and nothing registers on the global default any more.
/metrics is served from the new registry with the same series names, labels and auth. A test builds two metrics-enabled routers in one process.
Model: opus-5-5
upaas uploads its clone as a tar context, which .dockerignore does not filter, so its builds already carried .git; the unknown default of the VERSION build arg is what stamped them. The image now derives its version itself: ARG VERSION has no default, so make build falls back to script/version, which runs git describe on the copied .git; a VERSION build arg still wins.
.dockerignore sends .git without its config in a directory context and no longer leaves out tracked files, which would mark the tree -dirty. The builder installs git, trusts /build as a safe.directory, and fails when .git is present but the version comes out unknown. A shallow single-branch clone stamps its short commit.
Model: opus-5-5
Every 400, 403, 404 and 500 on an admin page now answers with an error page in the normal layout: the status, one fixed line explaining it, and a link back to the webhook list, or to sign-in when nobody is signed in. Unknown paths reach it through the router's not-found handler, a refused form token through the CSRF middleware, and a panic through a recoverer each admin page route group installs first. Status codes are unchanged, and the page always sends Cache-Control: no-store.
If the error page fails to render, the answer is the same status in plain text; if it panics, the answer is a 500. The receiver, the healthcheck and /metrics keep their plain answers.
Model: opus-5-5
Refusing an http or slack target whose address is private or reserved, on add or edit, now adds one sentence: such addresses are refused by default, and the server's ALLOWED_EGRESS_CIDRS setting allows named networks, with a pointer to the README section. It suggests no value, so it never points at allowing everything.
Metadata refusals get no such sentence. To tell them apart, the default blocklist's public addresses now have their own list, blockedPublicNetworks, still checked after the allowlist; a test pins which addresses each list refuses and how listing opens them, unchanged from before.
Model: opus-5-5
A logged-out GET of an admin page now redirects to /pages/login with its path and query in a next parameter, when they fit in 2048 bytes. The login form carries next as a hidden field; a successful login redirects there, and a failed one shows the page again with the same next. A POST still redirects to plain /pages/login.
The value is client-chosen, so every read of it goes through one check: after percent-decoding it must start with exactly one / and contain no backslash or control character; anything else becomes /. gosec's open-redirect rule is suppressed on the two redirects that follow it. The login page's navigation bar no longer links to itself.
Model: opus-5-5
The webhook page opens with a statistics pane: entrypoints and targets, deliveries in progress, the last arrival, the retention period; events, deliveries and failures, lifetime and within retention; and events, failures and failure percentage over 10 minutes and 24 hours.
Running totals, one row per target plus one for events, sit in each webhook's event database and are written in the same transaction as the rows they count; retention prunes in paused, stoppable batches and subtracts what it removes. Window figures are index-range counts on a new finished_at column. An existing event database must be recreated.
Model: opus-5-5
The webhook page and everything under it move from /source/ID to /hook/ID; the list and new-webhook form from /sources to /hooks and /hooks/new; the event log from .../logs to /hook/ID/events; entrypoint URLs, the ones senders post to, from /webhook/UUID to /h/UUID. The old paths are simply gone, per the owner's ruling. Every link, redirect, form action, test, comment and README mention follows; every link to the event log, and its title and heading, read "Full Event Log".
Keeping the UUID out of logs, metrics labels and error reports, and the receiver rate limits, key on the route itself, so they cover /h/ unchanged. Routed tests follow every link and form the pages render to a moved page.
Model: opus-5-5
Every test that starts a database seeds the admin account, hashing its password with Argon2id at 64 MB, about 150 MB under -race, and internal/handlers and internal/database ran dozens of those at once. That was the memory, and most of internal/handlers' run time; the product code does not leak.
HashPassword now hashes at a 1 MB cost only when testing.Testing() reports a test binary, so every test package gets it with nothing to add and a binary built by go build always hashes at the shipped parameters. TestHashPassword_ShippedParameters still hashes and verifies through HashPassword at the shipped cost. script/test adds -p 4 -parallel 8.
Model: opus-5-5
Unset or empty, TRUSTED_PROXIES now defaults to 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16, so a reverse proxy reaching webhooker from one of those ranges gets per-client rate-limit buckets with nothing set. A set value replaces the default entirely; an unparseable one still fails startup. The startup warning for an empty list is gone.
The README gives the default, one rule (if any client can reach webhooker, or the proxy in front of it, from an RFC 1918 source address, set the list to the proxy's address alone), and where to find that address: the remoteIP field of the http request log line. Loopback is not in the default.
Model: opus-5-5
The README's egress section and the comment above blockedNetworks now state what the default blocklist covers: the IPv4 private and reserved ranges, IPv6 loopback, unique local and link-local addresses, and certain public addresses, each added only because it hands credentials, user data or bootstrap material to whatever can reach it without the caller presenting anything. That is the same material as the rule above alwaysBlockedNetworks, so a future candidate can be accepted or refused against one rule.
A provider's other public addresses, such as 161.26.0.0/16 and 166.8.0.0/14, are not refused. Docs and a comment only; the lists are unchanged.
Model: opus-5-5
The webhook page's list, now headed "50 Most Recent Events", shows the 50 newest events, limited in the query. Each row adds the time received (relative, UTC on hover), the body size, the processing time (slowest delivery, queued to final outcome, retries included, "in progress" while pending), and, when the webhook has exactly one HTTP target, that target's latest HTTP status, coloured by class.
Each event now records its body size in a new body_bytes column when it is stored, so the list never reads bodies; the list loads only the delivery and attempt columns it shows, and a failed load is an error, not an empty list.
An existing event database must be recreated: the column is added in place, with no migration.
Model: opus-5-5
The profile page no longer shows the fixed "Account Type: Standard User" row. webhooker has one account, the administrator it creates, and no account types, so the row was untrue. The Account Information section now lists the username alone, and a test checks the row stays gone.
Model: opus-5-5
The build no longer downloads Alpine.js. Its npm package tarball is committed as 3p/alpinejs-3.14.9.tgz, byte for byte the file script/fetch-assets downloaded, with the sha256 that script pinned. script/assets (make assets) extracts package/dist/cdn.min.js to the ignored static/js/alpine.min.js; script/test runs it, so make test, make check, the pre-commit hook and the Dockerfile get the file with no network access, and make build, run and dev run it too.
Removed: script/fetch-assets, its Dockerfile step, static/vendor.sha256 and static/vendor_test.go. The README describes the new flow.
Model: opus-5-5
Usernames are limited to 1024 bytes, so no account can exist that is unable to log in. The username rides in the session cookie, which browsers and securecookie refuse past about 4 KB, leaving room for roughly 2000 bytes of username; the limit is about half that.
User.BeforeSave returns ErrUsernameTooLong when a whole User is created or saved. A byte-counting check constraint on users.username catches every other write, including a column update. The limit appears in the constant and in the struct tag; a test fails if they disagree.
Model: opus-5-5
A new database brings a new session key. A browser still holding the
old session cookie got a 500 on a correct login: Session.Get returned
the cookie's decode error and the login handler answered it with a
500. Get now treats a cookie that does not decode as absent, and
logging in replaces it. gorilla/csrf already did the same for the
CSRF cookie.
A start that creates webhooker.db now logs "created a new, empty
database" at WARN with its path, shortly before the first-boot banner,
so an unexpectedly empty DATA_DIR is noticed.
The codec tests now decode through the store, since Get no longer
reports the codec's reason.
Model: opus-5-5
The static file server was attached with Mount, which registers every
method, so POST, PUT and DELETE on an asset were answered 200 with the
file. It is now registered for GET and HEAD only, inside a /s group
whose method-not-allowed handler answers 405 with Allow: GET, HEAD.
A method chi does not route at all, such as PROPFIND, still gets 405
from the top-level router, without Allow. The inverted test and the
README route table say the same.
Model: opus-5-5
When a handler sets a cookie and then panics before sending
anything, the recover middleware now deletes Set-Cookie before
writing its 500, so a request that failed never hands the client
a credential. Every other header, Location included, is left as
http.Error leaves it, matching chi's Recoverer. A response that
was already sent is untouched.
Tests cover the uncommitted case (no cookie, Location kept) and
assert the cookie still reaches the client when the response was
committed before the panic.
Model: opus-5-5
Add 168.63.129.16 (Azure WireServer) to blockedNetworks, the default
blocklist, not alwaysBlockedNetworks: it is public unicast, so an
operator who lists it in ALLOWED_EGRESS_CIDRS can reach it again. The
refusal message, the allowlist startup warning, the README and the
comments no longer call every blocked address private/reserved, and
no longer claim the allowlist cannot open any metadata endpoint.
Sources:
- https://learn.microsoft.com/en-us/azure/virtual-network/what-is-ip-address-168-63-129-16
- https://learn.microsoft.com/en-us/azure/virtual-machines/metadata-security-protocol/overview
Deviation: 147.75.207.243 (Equinix Metal) is not added; Equinix
documents only a hostname, and the service was sunset on 2026-06-30.
Model: opus-5-5
The engine cached archive writers and never closed them at shutdown,
so after a clean stop an archive's rows could sit in its -wal while
the .db held no table. The engine's stop hook now evicts every cached
writer once its workers have returned, the same way deleting a webhook
does, so a clean stop leaves each archive as one file and a late write
is refused. If the workers do not return within the stop budget, the
writers are left open as a kill would leave them: closing would wait
on a write in progress, and a still-running worker would open new
ones.
The README no longer says archives keep their sidecars across a clean
stop.
Model: opus-5-5
A delivery set Content-Type from the event's ContentType and then
added the inbound Content-Type from the event's stored headers, so a
target could receive two values. The inbound Content-Type is no
longer forwarded from the stored headers; the receiver already saves
it as the event's ContentType.
Which value is sent is now stated at applyRequestHeaders: a
Content-Type configured on the target, otherwise the event's
ContentType, otherwise none. A configured one still survives a
cross-origin 307/308 with its body.
Model: opus-5-5
Restart recovery and the pending sweep skipped a pending delivery
whose target was missing from the batch's target map, every minute,
for the life of the database. A miss now asks loadTarget: no row
fails the delivery terminally with a recorded reason; any other error
leaves it pending, since the map is also empty when its query failed;
a target found there is used.
The failure goes through the ownership-gated function the retrying
paths already used, now failMissingTarget. Once it owns the delivery
it re-reads the row and fails it only if the status is unchanged, so
a delivery sent and settled in between is left alone.
Model: opus-5-5
While the breaker was half-open, Allow refused every delivery but the
probe and CooldownRemaining returned zero, so each queued task for the
target went straight back onto the retry channel and rewrote its status
on every pass until the probe finished.
CooldownRemaining now returns the whole cooldown while half-open, so a
refused delivery waits that long. A refused delivery already at
retrying is not written again, so the retry counter now moves only
when a refusal moves a delivery into retrying.
Model: opus-5-5
GetDB opened the database on a cache miss and then tried to cache it,
so callers racing on a webhook's first use could each open the file,
and the losers closed their copies. On a new file the parallel opens
also create its tables at the same time, and one caller can fail with
"table already exists".
A mutex now covers the open: GetDB looks in the cache again under it,
then opens and caches. DeleteDB and CloseAll take the same mutex, so
neither runs while an open is under way. Reading an already cached
database takes no lock.
The new test starts many callers on one webhook at once and checks
that exactly one open happened.
Model: opus-5-5
Restart recovery could find a just-written delivery pending, send it
and release it before the receiver's Notify queued the same delivery.
Notify's claim then succeeded on the released id, and the worker sent
it again because the new-task path never read the delivery's row.
Before sending a new task the worker now reads the delivery's status
by primary key and skips the task unless the row still says pending,
as the retry path already does for retrying. Nothing else can change
the row while the worker owns the delivery. A row left pending by a
failed bookkeeping write is still sent again.
loadRetryDelivery is renamed loadDelivery now that both paths use it.
Model: opus-5-5
The per-webhook event databases had no secondary indexes, so startup
recovery, the retry and pending sweeps, the queue-depth sampler, the
event log and retention each read whole tables. Indexes declared in
the GORM model tags now serve them, and AutoMigrate adds them to new
and existing databases alike.
Each index also covers deleted_at: GORM adds deleted_at IS NULL to
these queries, and SQLite, with no table statistics, otherwise
prefers the existing deleted_at index. A test checks SQLite's plan
for each statement as GORM builds it.
Rule suppressed: lll on the three event-tier model structs, whose
struct tags cannot wrap.
The resubmitted_from_id scan is left to
#325.
Model: opus-4-8 (implementation); opus-5-5 (rework)
sneak
clawbot