Render admin page errors in the normal layout (closes #382)

Every 400, 403, 404 and 500 on an admin page now answers with an
error page in the normal layout: one fixed line for the status and a
link back to the webhook list, or to sign-in when nobody is signed
in. The router's handler for unknown paths, the CSRF middleware's
refusal and a panic in an admin page route group use the same page;
each such group has its own recoverer and error reporting for that.
The page always sends Cache-Control: no-store. Status codes are
unchanged. The receiver, the healthcheck and /metrics keep their
plain answers. If the error page cannot render, or panics, the answer
is the same status in plain text.

Model: opus-5-5

README.md

@@ -142,7 +142,7 @@ TTY detection, and security headers are always applied.
 | `RETENTION_SWEEP_INTERVAL` | How often the retention reaper and archive sweeper run (Go duration, must be positive) | `1h` |
 | `SESSION_IDLE_TIMEOUT`  | Idle session timeout (Go duration)  | `24h`    |
 | `RECEIVER_RATE_LIMIT`   | Receiver requests/minute per IP per entrypoint (10x that per IP across the route) | `120` |
-| `TRUSTED_PROXIES`       | CIDRs whose forwarded headers are trusted (unset: all clients behind a proxy share one rate-limit bucket; a correct login password is never throttled either way) | `""` (none) |
+| `TRUSTED_PROXIES`       | CIDRs whose forwarded headers are trusted. A set value replaces the default. If any client can reach webhooker, or the proxy in front of it, from an RFC 1918 source address, set it to the proxy's address alone. See [Trusted proxies](#trusted-proxies) | `10.0.0.0/8,172.16.0.0/12,192.168.0.0/16` (RFC 1918) |
 | `ALLOWED_EGRESS_CIDRS`  | CIDRs that delivery targets may reach despite the SSRF blocklist. Read [Allowing egress to your own network](#allowing-egress-to-your-own-network) before setting it | `""` (none) |
 
 #### Allowing egress to your own network
@@ -389,41 +389,37 @@ unlocked.
 `TRUSTED_PROXIES` is a comma-separated list of CIDR blocks (a bare
 address such as `192.168.1.7` is accepted and treated as a single
 host), for example `192.168.1.7, 2001:db8::5`. It decides whose
-`X-Forwarded-For` header the rate limiters believe, so it should name
+`X-Forwarded-For` header the rate limiters believe, so it should cover
-the addresses of your reverse proxies and nothing else.
+the addresses of your reverse proxies.
 
 `X-Forwarded-For` is honoured **only** when the connecting peer is
 inside one of these blocks; for every other peer the client identity is
-the connection's own address and the header is ignored. The default is
+the connection's own address and the header is ignored. Unset (or
-the empty list, which trusts nobody — anything else would let any
+empty), the list is the RFC 1918 private ranges: `10.0.0.0/8`,
-client pick its own rate limit bucket, minting a fresh one per request
+`172.16.0.0/12` and `192.168.0.0/16`. A set value replaces the default
-or draining someone else's. Set it to the address of your reverse
+entirely. A set but unparseable value aborts startup.
-proxy, and to nothing wider. A set but unparseable value aborts
-startup.
 
-That default is safe against forged headers, but leaving it unset in
+If any client can reach webhooker, or the proxy in front of it, from an
-production has a cost you must know about. Production runs behind a
+RFC 1918 source address (directly, or through anything that can
-TLS-terminating reverse proxy, so with `TRUSTED_PROXIES` unset every
+rewrite source addresses, such as NAT or a published container port),
-request keys on the proxy's own address and all clients share a single
+set `TRUSTED_PROXIES` to the proxy's address alone, or every rate
-bucket per limit. The receiver limits become service-wide ceilings,
+limit, the webhook receiver's included, can be bypassed by those
-and the login endpoint's failure counting collapses onto one key, so a
+clients. The address to set is the `remoteIP` field of the
-stranger's wrong passwords throttle every other client's wrong
+`http request` log line for a request that came through the proxy.
-passwords.
+
+Behind a proxy the list does not cover, every request keys on the
+proxy's own address and all clients share a single bucket per limit.
+The receiver limits become service-wide ceilings, and the login
+endpoint's failure counting collapses onto one key, so a stranger's
+wrong passwords throttle every other client's wrong passwords. Set
+`TRUSTED_PROXIES` to that proxy's address to restore per-client
+buckets.
 
 What it cannot do is lock the operator out. The login endpoint
 verifies credentials **before** it consults any limit and charges only
 failures, so a correct password is never throttled no matter how full
 the bucket is. See [Rate Limiting](#rate-limiting).
 
-The remedy is to set `TRUSTED_PROXIES` to your reverse proxy's
-address, which restores per-client buckets. webhooker logs a warning
-at startup whenever `TRUSTED_PROXIES` is empty, in every environment,
-because behind a proxy every client shares one bucket in `dev` and
-`prod` alike. The warning is informational when nothing proxies to the
-process: with no proxy in front, the peer address is the client's own
-and the buckets are already per-client. See
-[Rate Limiting](#rate-limiting) for what each limit shares.
-
 `X-Real-IP` and `True-Client-IP` are **never** read, from any peer.
 Reverse proxies append to `X-Forwarded-For` but forward other client
 headers verbatim, so a single-valued header is client-controlled even
@@ -439,20 +435,10 @@ instead, since past such an entry the chain is not the shape assumed
 here. The peer address is likewise used when the header is absent or
 every hop in it is a trusted proxy.
 
-Two operator requirements follow:
+Your proxy must therefore **append** the peer address to
-
+`X-Forwarded-For` (nginx `$proxy_add_x_forwarded_for`, HAProxy
-- Your proxy must **append** the peer address to `X-Forwarded-For`
+`option forwardfor`, Caddy and AWS ALB by default), and must append a
-  (nginx `$proxy_add_x_forwarded_for`, HAProxy `option forwardfor`,
+bare address with no port.
-  Caddy and AWS ALB by default), and must append a bare address with
-  no port.
-- List proxy hosts **only**. Any address inside `TRUSTED_PROXIES`
-  chooses its own rate-limit key: its `X-Forwarded-For` is walked, so
-  it can name a different address on every request to get a fresh
-  bucket each time, or name another client's address to drain that
-  client's bucket. Never list a block that also covers clients — a
-  broad `10.0.0.0/8` on a network where clients live in the same range
-  makes all three limits, including the unauthenticated webhook
-  receiver, silently bypassable by every client in the block.
 
 #### Sessions
 
@@ -751,10 +737,15 @@ repository's `Dockerfile` and runs it. The app needs:
 - **Volume:** one host directory mounted at `/var/lib/webhooker`.
 - **Environment variables:**
     - `WEBHOOKER_ENVIRONMENT=prod`
-    - `TRUSTED_PROXIES`: your reverse proxy's address on that Docker
+    - `TRUSTED_PROXIES`: unset, it is the RFC 1918 ranges. Set it to
-      network. The `remoteIP` field of the `http request` log line for a
+      your reverse proxy's address alone if that address is outside
-      request that came through the proxy shows it; the health check's
+      those ranges, or if any client can reach webhooker, or the proxy,
-      own lines show `::1`. See [Trusted proxies](#trusted-proxies).
+      from an RFC 1918 source address (directly, or through anything
+      that can rewrite source addresses, such as NAT or a published
+      container port). The `remoteIP` field of the `http request` log
+      line for a request that came through the proxy shows that
+      address; the health check's own lines show `::1`. See
+      [Trusted proxies](#trusted-proxies).
     - Leave `BIND_ADDRESS` and `DATA_DIR` unset: the image sets
       `BIND_ADDRESS` to `0.0.0.0`, and `DATA_DIR` defaults to
       `/var/lib/webhooker`.
@@ -817,12 +808,16 @@ reports.
    behind a proxy means the `X-Forwarded-Proto` header. The block below
    sets it; without it every request is read as plaintext and cookies
    ship without `Secure`. See [Configuration](#configuration).
-3. **Set `TRUSTED_PROXIES` to the proxy's address.** Unset, every rate
+3. **Make sure `TRUSTED_PROXIES` covers the proxy's address.** For a
-   limiter keys on the connecting peer, which behind a proxy is the
+   proxy it does not cover, every rate limiter keys on the proxy, so
-   proxy on every request: all clients collapse into one global bucket
+   all clients share one bucket per limit. Unset, the list is the RFC
-   per limit and the receiver's per-IP limits become service-wide
+   1918 ranges, which do not cover a proxy that reaches the binary
-   ceilings. See [Trusted proxies](#trusted-proxies). List the proxy
+   itself over loopback (the binary bound to `127.0.0.1`). With the
-   and nothing else.
+   image, the address to check is the `remoteIP` field of the
+   `http request` log line for a request that came through the proxy.
+   If any client can reach webhooker, or the proxy, from an RFC 1918
+   source address, set the list to the proxy's address alone. See
+   [Trusted proxies](#trusted-proxies).
 4. **Send `Host` as `$http_host`, not `$host`.** `$host` strips the
    port. webhooker's Origin/Referer check compares against the host it
    was given, so on any port other than 443 `$host` makes every form
@@ -1375,10 +1370,11 @@ It uses:
 - **[go-chi/httprate](https://github.com/go-chi/httprate)** for
   sliding-window rate limiting of the password-change and webhook
   receiver endpoints. The bucket is per client IP only when
-  `TRUSTED_PROXIES` names the reverse proxy; unset, every client
+  `TRUSTED_PROXIES` covers the reverse proxy (by default it covers the
-  behind that proxy shares one bucket per limit. The login endpoint
+  RFC 1918 private ranges); otherwise every client behind that proxy
-  counts failed attempts itself instead, so that a correct password is
+  shares one bucket per limit. The login endpoint counts failed
-  never throttled (see [Rate Limiting](#rate-limiting))
+  attempts itself instead, so that a correct password is never
+  throttled (see [Rate Limiting](#rate-limiting))
 - **[Prometheus](https://prometheus.io)** for metrics, served at
   `/metrics` behind basic auth
 - **[Sentry](https://sentry.io)** for optional error reporting
@@ -2549,47 +2545,44 @@ the tree is checked out: four checkouts have reported 3,959, 3,961,
 client-supplied field was cut, and that the shipped chain's stack
 arrived uncut — never the numbers.
 
-Every limiter here — receiver, login, and password change — identifies
+Every limiter here — receiver, login, password change, delivery replay
-the client the same way, through one shared key function: the
+and event resubmit — identifies the client the same way, through one
-connection's own address, unless the peer is listed in
+shared key function: the connection's own address, unless the peer is
-`TRUSTED_PROXIES`, in which case the forwarded client address is used
+inside `TRUSTED_PROXIES`, in which case the forwarded client address is
-instead. That address becomes a bucket by family: IPv4 keys on the full
+used instead. That address becomes a bucket by family: IPv4 keys on
-address, IPv6 on its `/64` prefix. A routed `/64` is the normal
+the full address, IPv6 on its `/64` prefix. A routed `/64` is the normal
 residential and mobile IPv6 allocation, so keying IPv6 per address would
 let one subscriber rotate source addresses and mint a fresh bucket per
 request, evading these limits at the network layer without spoofing
 anything; the cost is that distinct clients inside one `/64` share a
 bucket. IPv4-mapped addresses (`::ffff:1.2.3.4`) key as the IPv4 address
-they carry. See [Trusted proxies](#trusted-proxies). Deployed without that
+they carry. See [Trusted proxies](#trusted-proxies). When that variable
-variable set, a client behind a reverse proxy shares one bucket with
+does not cover the reverse proxy, a client behind it shares one bucket
-every other client behind the same proxy. Set `TRUSTED_PROXIES` to the
+with every other client behind the same proxy. Set `TRUSTED_PROXIES` to
-proxy's address to get per-client limits back. What the shared bucket
+the proxy's address to get per-client limits back. What the shared bucket
 costs is not the same for every limiter, and the two cases pull in
 opposite directions:
 
 - For the **receiver** limits it costs throughput, which is the safe
   direction to be wrong in: sharing can only make a limit bind sooner,
   never let a sender past it. It matters more for the aggregate limit
-  than for the per-entrypoint one: with `TRUSTED_PROXIES` unset behind
+  than for the per-entrypoint one: with every request keyed on the
-  the reverse proxy a production deployment is required to run behind,
+  proxy, the aggregate limit becomes a service-wide ceiling of 1200
-  every request keys on the proxy, so the aggregate limit becomes a
+  requests per minute across all senders and all entrypoints, where the
-  service-wide ceiling of 1200 requests per minute across all senders
+  per-entrypoint limit's capacity still grows with the number of
-  and all entrypoints, where the per-entrypoint limit's capacity still
+  entrypoints.
-  grows with the number of entrypoints. Any deployment with more than a
-  handful of busy entrypoints must set `TRUSTED_PROXIES`.
 - For the **login and password-change** limits it costs precision, not
   availability. Login failures from every client land in one counter,
   so a stranger's wrong passwords make the operator's own wrong
   passwords answer `429` sooner; the operator's _correct_ password is
-  never affected, because it is never counted. Production deployments
+  never affected, because it is never counted.
-  should still set `TRUSTED_PROXIES`; webhooker warns at startup
-  whenever it is empty, in any environment.
 
 #### The login endpoint
 
 The login `POST` is the one endpoint with no pre-emptive limiter in
 front of it, and that is deliberate. A limiter that spends budget on
-arrival is a lockout in this deployment shape: sharing one bucket, a
+arrival is a lockout wherever clients share one bucket, as they do
+behind a reverse proxy that `TRUSTED_PROXIES` does not cover: a
 stranger sending five POSTs a minute — about 0.08 requests per second,
 from anywhere — keeps it permanently full, and the operator has no
 second administrative path. So the handler inverts the order:
@@ -2686,8 +2679,10 @@ re-fills both verification slots on its first two requests. The
 remedies are to block the source at the reverse proxy, or to
 rate-limit `POST /pages/login` there — the one place a limit can be
 applied without reintroducing the lockout, because the proxy sees the
-real client address. Setting `TRUSTED_PROXIES` does not stop the
+real client address. `TRUSTED_PROXIES` does not stop the saturation.
-saturation, but it makes the source visible in the failure logs.
+The flood's source is in the proxy's access log: webhooker's own logs
+record the proxy's address, not the client's (see
+[Deployment behind a reverse proxy](#deployment-behind-a-reverse-proxy)).
 
 Finer-grained per-webhook rate limits (configured in the web UI and
 enforced in the webhook handler) can layer on top of this env-level
@@ -3045,10 +3040,9 @@ check, see [The login endpoint](#the-login-endpoint).
   It runs behind session auth, so only a client already holding a
   valid session reaches it, and an operator throttled out of changing
   a password can still log in. The bucket is per client IP only when
-  `TRUSTED_PROXIES` names the reverse proxy; unset, every client
+  `TRUSTED_PROXIES` covers the reverse proxy; otherwise every client
   shares one bucket, which costs precision rather than availability
-  (see [Rate Limiting](#rate-limiting)). webhooker warns at startup
+  (see [Rate Limiting](#rate-limiting))
-  whenever `TRUSTED_PROXIES` is empty
 - Prometheus metrics behind basic auth
 - Static assets embedded in binary (no filesystem access needed at
   runtime)

TODO.md

@@ -387,7 +387,7 @@ point of the branch.
 - 2026-03-05 security headers middleware, session regeneration on
   login, request body size limits (#41)
 - 2026-03-04 tests for delivery, middleware, and session packages
-  (#32); removed globals.Buildarch (#31)
+  (#32); removed the build-architecture global (#31)
 - 2026-03-04 1.0 MVP merge: Webhook/Entrypoint/Target rename, core
   delivery engine with bounded worker pool and circuit breaker,
   parallel fan-out, per-webhook event databases, management UI (#16)

internal/config/config.go

@@ -75,6 +75,11 @@ const (
 	// internet-exposed endpoint.
 	defaultReceiverRateLimit = 120
 
+	// defaultTrustedProxies is TRUSTED_PROXIES when it is unset: the
+	// RFC 1918 private ranges, which a reverse proxy reaching the
+	// process over a Docker network or a private LAN connects from.
+	defaultTrustedProxies = "10.0.0.0/8,172.16.0.0/12,192.168.0.0/16"
+
 	// maxPort is the highest valid TCP port number. The lower
 	// bound (at least 1) is enforced by envPositiveInt.
 	maxPort = 65535
@@ -172,13 +177,14 @@ type Config struct {
 
 	// TrustedProxies is the set of networks whose members are
 	// allowed to speak for the client with X-Forwarded-For, the
-	// only forwarded header read. It is empty unless
+	// only forwarded header read. Unless TRUSTED_PROXIES is set it
-	// TRUSTED_PROXIES is set, and empty means no peer is
+	// is the RFC 1918 private ranges (defaultTrustedProxies); a set
-	// trusted: forwarded headers are then ignored entirely and
+	// value replaces them. If any client can reach the process, or
-	// clients are identified by the connection's own address.
+	// the proxy in front of it, from an RFC 1918 source address
-	// Members can choose their own rate-limit key, so this must
+	// (directly, or through anything that can rewrite source
-	// name proxy hosts only, never a block that also covers
+	// addresses, such as NAT or a published container port), it
-	// clients.
+	// must be set to the proxy's address alone, or every rate limit
+	// can be bypassed by those clients.
 	TrustedProxies []netip.Prefix
 
 	// AllowedEgressCIDRs is the set of networks a delivery target
@@ -460,14 +466,15 @@ func parseCIDR(entry string) (netip.Prefix, error) {
 
 // envPrefixList returns the value of the named environment variable
 // parsed as a comma-separated list of CIDR blocks (bare addresses
-// allowed). An unset, empty, or blank value yields an empty list. A
+// allowed). An unset, empty, or blank value is read as defaultValue
-// set value containing an unparseable entry is a hard error naming
+// instead. A set value containing an unparseable entry is a hard
-// the key and the bad entry, so startup fails loudly rather than
+// error naming the key and the bad entry, so startup fails loudly
-// silently running with a list the operator did not intend.
+// rather than silently running with a list the operator did not
-func envPrefixList(key string) ([]netip.Prefix, error) {
+// intend.
+func envPrefixList(key, defaultValue string) ([]netip.Prefix, error) {
 	v := strings.TrimSpace(os.Getenv(key))
 	if v == "" {
-		return nil, nil
+		v = defaultValue
 	}
 
 	var prefixes []netip.Prefix
@@ -681,12 +688,12 @@ func loadFromEnv() (*Config, error) {
 		return nil, err
 	}
 
-	trustedProxies, err := envPrefixList("TRUSTED_PROXIES")
+	trustedProxies, err := envPrefixList("TRUSTED_PROXIES", defaultTrustedProxies)
 	if err != nil {
 		return nil, err
 	}
 
-	allowedEgressCIDRs, err := envPrefixList("ALLOWED_EGRESS_CIDRS")
+	allowedEgressCIDRs, err := envPrefixList("ALLOWED_EGRESS_CIDRS", "")
 	if err != nil {
 		return nil, err
 	}
@@ -760,50 +767,6 @@ func (c *Config) warnEgressAllowlist(log *slog.Logger) {
 	)
 }
 
-// warnSharedRateLimitBucket logs a startup warning whenever
-// TRUSTED_PROXIES is empty, in any environment.
-//
-// With no trusted proxies every rate limiter keys on the connecting
-// peer's address. Whether that is harmless or dangerous depends on
-// what is in front of the process, which this code cannot observe:
-// with nothing in front, the peer is the client and the limits are
-// per-client as intended; behind a reverse proxy the peer is the proxy
-// for every request, so all clients share one bucket per limiter.
-//
-// The login endpoint no longer spends budget on arrival — it verifies
-// credentials first and charges only failures — so a shared bucket
-// cannot deny the operator a correct password. What it does collapse
-// is the failure counting: one client's wrong passwords throttle
-// everyone else's wrong passwords, and the receiver's limits become
-// service-wide ceilings.
-//
-// The warning is deliberately not gated on WEBHOOKER_ENVIRONMENT:
-// behind a proxy every client shares one bucket in dev and prod alike.
-//
-// The default of trusting nobody is deliberate — trusting forwarded
-// headers from arbitrary peers lets any client choose its own bucket —
-// so this warns rather than failing startup or changing the key.
-func (c *Config) warnSharedRateLimitBucket(log *slog.Logger) {
-	if len(c.TrustedProxies) > 0 {
-		return
-	}
-
-	log.Warn(
-		"TRUSTED_PROXIES is empty: every rate limit keys on the "+
-			"connecting peer's address. With nothing proxying to "+
-			"this process that is the client itself and the limits "+
-			"are per-client as intended. Behind a reverse proxy the "+
-			"peer is the proxy on every request, so all clients "+
-			"share one bucket per limit: the receiver limits become "+
-			"service-wide ceilings, and one client's failed logins "+
-			"throttle every other client's failed logins — a "+
-			"correct password still gets in. If anything proxies to "+
-			"this process, set TRUSTED_PROXIES to its address.",
-		"environment", c.Environment,
-		"trustedProxies", len(c.TrustedProxies),
-	)
-}
-
 // New creates a Config by reading environment variables.
 //
 //nolint:revive // lc parameter is required by fx even if unused.
@@ -849,7 +812,6 @@ func New(lc fx.Lifecycle, params ConfigParams) (*Config, error) {
 		"hasMetricsAuth", s.MetricsAuthEnabled(),
 	)
 
-	s.warnSharedRateLimitBucket(log)
 	s.warnEgressAllowlist(log)
 
 	return s, nil

internal/config/config_test.go

@@ -551,6 +551,11 @@ func testReceiverRateLimitSuccess(
 }
 
 func TestTrustedProxies(t *testing.T) {
+	// Unset, the RFC 1918 private ranges are trusted, so a reverse
+	// proxy on a Docker network or a private LAN is covered without
+	// configuration.
+	defaultProxies := []string{cidrPrivateV4, "172.16.0.0/12", "192.168.0.0/16"}
+
 	tests := []struct {
 		name        string
 		set         bool
@@ -559,18 +564,21 @@ func TestTrustedProxies(t *testing.T) {
 		expected    []string
 	}{
 		{
-			// The default must be "trust nobody": an empty list
-			// means forwarded headers are ignored, never that
-			// every peer may speak for the client.
 			name:     caseUnsetUsesDefault,
 			set:      false,
-			expected: []string{},
+			expected: defaultProxies,
 		},
 		{
-			name:     "blank value trusts nothing",
+			name:     "blank value uses default",
 			set:      true,
 			value:    "  ",
-			expected: []string{},
+			expected: defaultProxies,
+		},
+		{
+			name:     "set value replaces the default entirely",
+			set:      true,
+			value:    "203.0.113.7",
+			expected: []string{"203.0.113.7/32"},
 		},
 		{
 			name:  caseValidValueParsed,
@@ -845,101 +853,6 @@ func TestEgressAllowlistWarning(t *testing.T) {
 	}
 }
 
-// TestSharedRateLimitBucketWarning covers the startup warning that
-// tells an operator a deployment behind a reverse proxy shares one
-// rate-limit bucket between every client, which turns the receiver
-// limits into service-wide ceilings and collapses login failure
-// counting. It must fire whenever TRUSTED_PROXIES is empty, in any
-// environment, because behind a proxy every client shares one bucket
-// in dev and prod alike. It stays quiet once proxies are named.
-func TestSharedRateLimitBucketWarning(t *testing.T) {
-	tests := []struct {
-		name           string
-		environment    string
-		trustedProxies string
-		expectWarning  bool
-	}{
-		{
-			name:          "prod without trusted proxies warns",
-			environment:   config.EnvironmentProd,
-			expectWarning: true,
-		},
-		{
-			name:           "prod with trusted proxies is quiet",
-			environment:    config.EnvironmentProd,
-			trustedProxies: cidrPrivateV4,
-			expectWarning:  false,
-		},
-		{
-			name:          "dev without trusted proxies warns",
-			environment:   config.EnvironmentDev,
-			expectWarning: true,
-		},
-		{
-			name:           "dev with trusted proxies is quiet",
-			environment:    config.EnvironmentDev,
-			trustedProxies: cidrPrivateV4,
-			expectWarning:  false,
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			// Cannot use t.Parallel() here because t.Setenv
-			// is incompatible with parallel subtests.
-			t.Setenv("WEBHOOKER_ENVIRONMENT", tt.environment)
-
-			if tt.trustedProxies == "" {
-				require.NoError(
-					t, os.Unsetenv("TRUSTED_PROXIES"),
-				)
-			} else {
-				t.Setenv("TRUSTED_PROXIES", tt.trustedProxies)
-			}
-
-			var buf bytes.Buffer
-
-			log := slog.New(slog.NewJSONHandler(
-				&buf, &slog.HandlerOptions{
-					Level: slog.LevelDebug,
-				},
-			))
-
-			require.NoError(
-				t,
-				config.WarnSharedRateLimitBucketForTest(log),
-			)
-
-			if !tt.expectWarning {
-				assert.Empty(t, buf.String())
-
-				return
-			}
-
-			logged := buf.String()
-
-			assert.Contains(t, logged, `"level":"WARN"`)
-			assert.Contains(t, logged, "TRUSTED_PROXIES")
-			assert.Contains(t, logged, "share one bucket")
-			assert.Contains(
-				t, logged, "throttle every other client's failed logins",
-			)
-			// The warning must not claim a lockout the login
-			// endpoint no longer permits: credentials are verified
-			// before any budget is spent.
-			assert.Contains(
-				t, logged, "a correct password still gets in",
-			)
-			// The text must stay accurate for a developer with
-			// nothing in front of the process, where an empty
-			// list costs nothing.
-			assert.Contains(
-				t, logged, "nothing proxying to this process",
-			)
-		})
-	}
-}
-
 // metricsEnv describes what one subtest below puts in the
 // environment for a single METRICS_ variable. A variable that is
 // set to the empty string and one that is not set at all are

internal/config/export_test.go

@@ -6,21 +6,6 @@ import "log/slog"
 // the external config_test package so each helper can be covered by
 // its own table-driven test without weakening the package API.
 
-// WarnSharedRateLimitBucketForTest loads a Config from the current
-// environment and emits its startup warnings to log. The real logger
-// writes to stdout, so this lets the warning's firing condition be
-// asserted against a handler the test controls.
-func WarnSharedRateLimitBucketForTest(log *slog.Logger) error {
-	c, err := loadFromEnv()
-	if err != nil {
-		return err
-	}
-
-	c.warnSharedRateLimitBucket(log)
-
-	return nil
-}
-
 // WarnEgressAllowlistForTest loads a Config from the current
 // environment and emits its egress-allowlist startup warning to
 // log, so a test can assert both that the warning fires only when

internal/handlers/auth.go

@@ -36,7 +36,7 @@ func (h *Handlers) HandleLoginSubmit() http.HandlerFunc {
 		err := r.ParseForm()
 		if err != nil {
 			h.log.Error("failed to parse form", "error", err)
-			http.Error(w, "Bad request", http.StatusBadRequest)
+			h.renderError(w, r, http.StatusBadRequest)
 
 			return
 		}
@@ -103,9 +103,10 @@ func (h *Handlers) renderLoginError(
 // The credential check runs BEFORE any rate-limit budget is
 // consulted, and only a failed check spends budget. That is what
 // keeps the single administrative path reachable: behind the reverse
-// proxy this deployment requires, with TRUSTED_PROXIES unset, every
+// proxy this deployment requires, when TRUSTED_PROXIES does not cover
-// client shares one bucket, so a limiter spent on arrival lets any
+// it, every client shares one bucket, so a limiter spent on arrival
-// stranger deny the operator's own correct password indefinitely.
+// lets any stranger deny the operator's own correct password
+// indefinitely.
 //
 // Verifying first means every login POST costs an Argon2id hash, so
 // the work is taken under a bounded number of verification slots.
@@ -165,11 +166,7 @@ func (h *Handlers) authenticateUser(
 
 	valid, err := database.VerifyPassword(password, user.Password)
 	if err != nil {
-		h.log.Error("failed to verify password", "error", err)
+		h.serverError(w, r, "failed to verify password", err)
-		http.Error(
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return user, err
 	}
@@ -241,24 +238,14 @@ func (h *Handlers) createAuthenticatedSession(
 ) error {
 	oldSess, err := h.session.Get(r)
 	if err != nil {
-		h.log.Error("failed to get session", "error", err)
+		h.serverError(w, r, "failed to get session", err)
-		http.Error(
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return err
 	}
 
 	sess, err := h.session.Regenerate(r, w, oldSess)
 	if err != nil {
-		h.log.Error(
+		h.serverError(w, r, "failed to regenerate session", err)
-			"failed to regenerate session", "error", err,
-		)
-		http.Error(
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return err
 	}
@@ -267,11 +254,7 @@ func (h *Handlers) createAuthenticatedSession(
 
 	err = h.session.Save(r, w, sess)
 	if err != nil {
-		h.log.Error("failed to save session", "error", err)
+		h.serverError(w, r, "failed to save session", err)
-		http.Error(
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return err
 	}

internal/handlers/auth_test.go

@@ -25,7 +25,7 @@ const (
 
 	// sharedProxyPeer is the whole point of this file. Production is
 	// required to run behind a TLS-terminating reverse proxy, and
-	// TRUSTED_PROXIES defaults to empty, so every client — attacker
+	// when TRUSTED_PROXIES does not cover it every client — attacker
 	// and operator alike — reaches the process from the proxy's
 	// address and shares one rate-limit bucket. Both parties in
 	// these tests therefore use the same RemoteAddr.
@@ -115,11 +115,11 @@ func floodFailures(
 // done-criterion of https://git.eeqj.de/sneak/webhooker/issues/150.
 //
 // The attacker and the operator share one rate-limit bucket, because
-// behind the mandated reverse proxy with TRUSTED_PROXIES unset every
+// behind the mandated reverse proxy, when TRUSTED_PROXIES does not
-// client keys on the proxy's address. The attacker floods the
+// cover it, every client keys on the proxy's address. The attacker
-// operator's own username — a single-admin product has a predictable
+// floods the operator's own username — a single-admin product has a
-// one — far past the failure limit. The operator must still be able
+// predictable one — far past the failure limit. The operator must
-// to log in with the correct password.
+// still be able to log in with the correct password.
 //
 // This fails if credentials stop being verified ahead of the limiter.
 func TestLogin_StrangersFloodCannotLockOutTheOperator(t *testing.T) {

internal/handlers/delivery_replay.go

@@ -105,9 +105,7 @@ func (h *Handlers) HandleDeliveryReplay() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err := r.ParseForm()
 		if err != nil {
-			http.Error(
+			h.renderError(w, r, http.StatusBadRequest)
-				w, "Bad request", http.StatusBadRequest,
-			)
 
 			return
 		}
@@ -124,14 +122,14 @@ func (h *Handlers) replayDelivery(
 	webhook database.Webhook,
 ) {
 	if !h.dbMgr.DBExists(webhook.ID) {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return
 	}
 
 	webhookDB, err := h.dbMgr.GetDB(webhook.ID)
 	if err != nil {
-		h.serverError(w, "failed to get webhook database", err)
+		h.serverError(w, r, "failed to get webhook database", err)
 
 		return
 	}
@@ -173,7 +171,7 @@ func (h *Handlers) loadReplaySource(
 			&original, "id = ?", chi.URLParam(r, "deliveryID"),
 		).Error
 	if err != nil {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return nil, false
 	}
@@ -195,7 +193,7 @@ func (h *Handlers) queueReplay(
 	)
 	if err != nil {
 		h.serverError(
-			w, "failed to count in-flight deliveries", err,
+			w, r, "failed to count in-flight deliveries", err,
 		)
 
 		return
@@ -212,7 +210,7 @@ func (h *Handlers) queueReplay(
 	err = webhookDB.
 		First(&event, "id = ?", original.EventID).Error
 	if err != nil {
-		h.serverError(w, "failed to load event for replay", err)
+		h.serverError(w, r, "failed to load event for replay", err)
 
 		return
 	}
@@ -222,7 +220,7 @@ func (h *Handlers) queueReplay(
 	)
 	if err != nil {
 		h.serverError(
-			w, "failed to create replay delivery", err,
+			w, r, "failed to create replay delivery", err,
 		)
 
 		return

internal/handlers/error_page_test.go

@@ -0,0 +1,53 @@
+package handlers_test
+
+import (
+	"context"
+	"html/template"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"sneak.berlin/go/webhooker/internal/handlers"
+)
+
+// TestErrorPage_RenderFailureKeepsStatus proves that an error page
+// which cannot render answers with the status it was reporting, as
+// plain text, and is not attempted again: a page whose own render
+// fails reaches the error page, and the error page failing as well
+// ends there with the 500.
+func TestErrorPage_RenderFailureKeepsStatus(t *testing.T) {
+	t.Parallel()
+
+	var h *handlers.Handlers
+
+	app := newTestApp(t, &h)
+	app.RequireStart()
+
+	t.Cleanup(app.RequireStop)
+
+	// .Status is an int, so asking it for a field fails the render.
+	failing := `{{.Status.Missing}}`
+	h.AddTemplateForTest("error.html", template.Must(
+		template.New("error").Parse(failing),
+	))
+	h.AddTemplateForTest("failing.html", template.Must(
+		template.New("failing").Parse(`{{.Data.Missing}}`),
+	))
+
+	req := httptest.NewRequestWithContext(
+		context.Background(), http.MethodGet, "/", nil,
+	)
+
+	w := httptest.NewRecorder()
+	h.HandleErrorPage(http.StatusNotFound).ServeHTTP(w, req)
+
+	assert.Equal(t, http.StatusNotFound, w.Code)
+	assert.Equal(t, "Not Found\n", w.Body.String())
+
+	w = httptest.NewRecorder()
+	h.RenderTemplateForTest(w, req, "failing.html", 0)
+
+	assert.Equal(t, http.StatusInternalServerError, w.Code)
+	assert.Equal(t, "Internal Server Error\n", w.Body.String())
+}

internal/handlers/event_body.go

@@ -52,7 +52,7 @@ func (h *Handlers) HandleEventBodyDownload() http.HandlerFunc {
 		// steered by a client.
 		eventID, err := uuid.Parse(chi.URLParam(r, "eventID"))
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -103,21 +103,21 @@ func (h *Handlers) serveEventBody(
 	eventID string,
 ) {
 	if !h.dbMgr.DBExists(webhook.ID) {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return
 	}
 
 	webhookDB, err := h.dbMgr.GetDB(webhook.ID)
 	if err != nil {
-		h.serverError(w, "failed to get webhook database", err)
+		h.serverError(w, r, "failed to get webhook database", err)
 
 		return
 	}
 
 	body, found, err := eventBody(webhookDB, webhook.ID, eventID)
 	if err != nil {
-		h.serverError(w, "failed to read event body", err)
+		h.serverError(w, r, "failed to read event body", err)
 
 		return
 	}
@@ -130,7 +130,7 @@ func (h *Handlers) serveEventBody(
 	// row and the whole body is served, or it does not and the
 	// response is a clean 404.
 	if !found {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return
 	}

internal/handlers/event_resubmit.go

@@ -99,7 +99,7 @@ func (h *Handlers) HandleEventResubmit() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err := r.ParseForm()
 		if err != nil {
-			http.Error(w, "Bad request", http.StatusBadRequest)
+			h.renderError(w, r, http.StatusBadRequest)
 
 			return
 		}
@@ -120,20 +120,20 @@ func (h *Handlers) resubmitEvent(
 	// alphabet rather than from the request.
 	eventID, err := uuid.Parse(chi.URLParam(r, "eventID"))
 	if err != nil {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return
 	}
 
 	if !h.dbMgr.DBExists(webhook.ID) {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return
 	}
 
 	webhookDB, err := h.dbMgr.GetDB(webhook.ID)
 	if err != nil {
-		h.serverError(w, "failed to get webhook database", err)
+		h.serverError(w, r, "failed to get webhook database", err)
 
 		return
 	}
@@ -147,7 +147,7 @@ func (h *Handlers) resubmitEvent(
 		webhookDB, webhook.ID, eventID.String(),
 	)
 	if err != nil {
-		h.serverError(w, "failed to load event to resubmit", err)
+		h.serverError(w, r, "failed to load event to resubmit", err)
 
 		return
 	}
@@ -155,7 +155,7 @@ func (h *Handlers) resubmitEvent(
 	// A miss is a 404 whether the event was reaped, belongs to
 	// another webhook, or never existed.
 	if !found {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return
 	}
@@ -207,7 +207,7 @@ func (h *Handlers) queueResubmit(
 	// inactive one is skipped rather than refused.
 	targets, err := h.loadActiveTargets(webhook.ID)
 	if err != nil {
-		h.serverError(w, "failed to query targets", err)
+		h.serverError(w, r, "failed to query targets", err)
 
 		return
 	}
@@ -225,7 +225,7 @@ func (h *Handlers) queueResubmit(
 		targets,
 	)
 	if err != nil {
-		h.serverError(w, "failed to store resubmitted event", err)
+		h.serverError(w, r, "failed to store resubmitted event", err)
 
 		return
 	}

internal/handlers/export_test.go

@@ -1,9 +1,11 @@
 package handlers
 
 import (
+	"context"
 	"html/template"
 	"log/slog"
 	"net/http"
+	"net/http/httptest"
 
 	"sneak.berlin/go/webhooker/internal/database"
 )
@@ -63,12 +65,20 @@ func (s *Handlers) LoadEventLogViewsForTest(
 	page int,
 ) []EventLogView {
 	views, _, _ := s.loadEventsWithDeliveries(
-		w, webhook, nil, page,
+		w, newRequestForTest(), webhook, nil, page,
 	)
 
 	return views
 }
 
+// newRequestForTest is the request the helpers here pass on for
+// callers that have none: it is used only to render the error page.
+func newRequestForTest() *http.Request {
+	return httptest.NewRequestWithContext(
+		context.Background(), http.MethodGet, "/", nil,
+	)
+}
+
 // AddTemplateForTest registers a template under a page name so that
 // the handlers_test package can drive the render path with a
 // template of its own.
@@ -122,5 +132,5 @@ func (s *Handlers) BuildDatabaseTargetConfigForTest(
 	w http.ResponseWriter,
 	expiry string,
 ) (string, error) {
-	return s.buildDatabaseTargetConfig(w, expiry)
+	return s.buildDatabaseTargetConfig(w, newRequestForTest(), expiry)
 }

internal/handlers/handlers.go

@@ -135,6 +135,7 @@ func New(
 		"source_edit.html":   parsePageTemplate("source_edit.html"),
 		"source_logs.html":   parsePageTemplate("source_logs.html"),
 		"target_edit.html":   parsePageTemplate("target_edit.html"),
+		"error.html":         parsePageTemplate("error.html"),
 	}
 
 	lc.Append(fx.Hook{
@@ -146,6 +147,15 @@ func New(
 	return s, nil
 }
 
+// HandleErrorPage returns a handler that answers every request with
+// the error page for status. The router uses it for unknown paths and
+// the CSRF middleware for a refused form.
+func (s *Handlers) HandleErrorPage(status int) http.HandlerFunc {
+	return func(w http.ResponseWriter, r *http.Request) {
+		s.renderError(w, r, status)
+	}
+}
+
 func (s *Handlers) respondJSON(
 	w http.ResponseWriter,
 	_ *http.Request,
@@ -163,15 +173,76 @@ func (s *Handlers) respondJSON(
 	}
 }
 
-// serverError logs an error and sends a 500 response.
+// serverError logs an error and answers with the 500 error page.
 func (s *Handlers) serverError(
-	w http.ResponseWriter, msg string, err error,
+	w http.ResponseWriter, r *http.Request, msg string, err error,
 ) {
 	s.log.Error(msg, "error", err)
-	http.Error(
+	s.renderError(w, r, http.StatusInternalServerError)
-		w, "Internal server error",
+}
-		http.StatusInternalServerError,
+
-	)
+// renderError answers with status and the error page: the normal
+// layout, one fixed line explaining the status, and a link back to the
+// webhook list, or to sign-in when nobody is signed in.
+//
+// It renders the page itself rather than through renderTemplate,
+// whose own failure comes here. If the error page cannot render
+// either, the answer is the same status in plain text: never a second
+// attempt, and never a different status.
+func (s *Handlers) renderError(
+	w http.ResponseWriter,
+	r *http.Request,
+	status int,
+) {
+	// The page names the signed-in user, and some error pages are
+	// served outside the routes where NoCache runs.
+	w.Header().Set("Cache-Control", "no-store")
+
+	data := s.pageData(r, map[string]any{
+		"Status":     status,
+		"StatusText": http.StatusText(status),
+		"Message":    errorPageText(status),
+	})
+
+	var buf bytes.Buffer
+
+	err := s.templates["error.html"].Execute(&buf, data)
+	if err != nil {
+		s.log.Error("failed to render error page", "error", err)
+		http.Error(w, http.StatusText(status), status)
+
+		return
+	}
+
+	w.Header().Set("Content-Type", "text/html; charset=utf-8")
+	w.WriteHeader(status)
+
+	_, err = buf.WriteTo(w)
+	if err != nil {
+		s.log.Error("failed to write error page", "error", err)
+	}
+}
+
+// errorPageText is the line the error page shows for status. It is
+// fixed per status, so the page tells the reader no more than the
+// plain-text answers it replaced did.
+func errorPageText(status int) string {
+	switch status {
+	case http.StatusBadRequest:
+		return "The request could not be read."
+	case http.StatusForbidden:
+		return "The request was refused. If it came from a form " +
+			"left open for a long time, reload the page and try " +
+			"again."
+	case http.StatusNotFound:
+		return "There is nothing here. It may have been deleted, " +
+			"or the address may be wrong."
+	case http.StatusServiceUnavailable:
+		return "The server is busy. Please try again in a moment."
+	default: // http.StatusInternalServerError
+		return "Something went wrong on the server. Please try " +
+			"again."
+	}
 }
 
 // UserInfo represents user information for templates
@@ -224,14 +295,17 @@ func (s *Handlers) renderTemplate(
 			"template not found",
 			"template", pageTemplate,
 		)
-		http.Error(
+		s.renderError(w, r, http.StatusInternalServerError)
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return
 	}
 
+	s.executeTemplate(w, r, tmpl, s.pageData(r, data))
+}
+
+// pageData adds the fields the shared layout renders to a page's own
+// data.
+func (s *Handlers) pageData(r *http.Request, data any) any {
 	userInfo := s.getUserInfo(r)
 	csrfToken := middleware.CSRFToken(r)
 
@@ -245,19 +319,16 @@ func (s *Handlers) renderTemplate(
 		m["User"] = userInfo
 		m["CSRFToken"] = csrfToken
 		m["Version"] = version
-		s.executeTemplate(w, tmpl, m)
 
-		return
+		return m
 	}
 
-	wrapper := templateDataWrapper{
+	return templateDataWrapper{
 		User:      userInfo,
 		CSRFToken: csrfToken,
 		Version:   version,
 		Data:      data,
 	}
-
-	s.executeTemplate(w, tmpl, wrapper)
 }
 
 // executeTemplate renders the template into a buffer and writes to
@@ -270,6 +341,7 @@ func (s *Handlers) renderTemplate(
 // this reason.
 func (s *Handlers) executeTemplate(
 	w http.ResponseWriter,
+	r *http.Request,
 	tmpl *template.Template,
 	data any,
 ) {
@@ -280,10 +352,7 @@ func (s *Handlers) executeTemplate(
 		s.log.Error(
 			"failed to execute template", "error", err,
 		)
-		http.Error(
+		s.renderError(w, r, http.StatusInternalServerError)
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return
 	}

internal/handlers/handlers_test.go

@@ -307,10 +307,14 @@ func TestRenderTemplateMidRenderErrorSendsNoPartialBody(t *testing.T) {
 		t, http.StatusInternalServerError, w.Code,
 		"a failed render must report a 500",
 	)
-	assert.Equal(
+	assert.NotContains(
-		t, "Internal server error\n", w.Body.String(),
+		t, w.Body.String(), partialPageMarker,
 		"the response must carry no part of the aborted page",
 	)
+	assert.Contains(
+		t, w.Body.String(), "500 Internal Server Error",
+		"a failed render must answer with the error page",
+	)
 }
 
 func TestBuildDatabaseTargetConfig_Valid(t *testing.T) {

internal/handlers/profile.go

@@ -1,7 +1,6 @@
 package handlers
 
 import (
-	"context"
 	"net/http"
 
 	"github.com/go-chi/chi"
@@ -37,14 +36,14 @@ func (h *Handlers) HandlePasswordChange() http.HandlerFunc {
 		err := r.ParseForm()
 		if err != nil {
 			h.log.Error("failed to parse form", "error", err)
-			http.Error(w, "Bad request", http.StatusBadRequest)
+			h.renderError(w, r, http.StatusBadRequest)
 
 			return
 		}
 
 		successMessage, errorMessage, handled := h.applyPasswordChange(
-			r.Context(),
 			w,
+			r,
 			sessionUsername,
 			// PostFormValue, not FormValue: the credential must
 			// come from the body, never from the query string.
@@ -66,12 +65,12 @@ func (h *Handlers) HandlePasswordChange() http.HandlerFunc {
 // applyPasswordChange verifies the current password and, on success,
 // persists a fresh hash for the user, reusing the same helpers that
 // bootstrap the admin user. It returns the success and error messages
-// to display on the profile page. On an internal failure it writes a
+// to display on the profile page. On an internal failure it writes the
-// 500 response itself and returns handled=false, signalling the caller
+// error page itself and returns handled=false, signalling the caller
 // to stop without re-rendering the page.
 func (h *Handlers) applyPasswordChange(
-	ctx context.Context,
 	w http.ResponseWriter,
+	r *http.Request,
 	username, currentPassword, newPassword, confirmPassword string,
 ) (string, string, bool) {
 	// This endpoint verifies one password and hashes another, at
@@ -79,15 +78,10 @@ func (h *Handlers) applyPasswordChange(
 	// endpoint uses. The bound is per hash, not per endpoint: leaving
 	// this path outside it would leave a hole in it. The slot is held
 	// across both hashes.
-	release, ok := h.mw.BeginPasswordVerification(ctx)
+	release, ok := h.mw.BeginPasswordVerification(r.Context())
 	if !ok {
 		h.log.Warn("password verification capacity exhausted")
-		http.Error(
+		h.renderError(w, r, http.StatusServiceUnavailable)
-			w,
-			"The server is busy verifying credentials. "+
-				"Please try again.",
-			http.StatusServiceUnavailable,
-		)
 
 		return "", "", false
 	}
@@ -103,7 +97,7 @@ func (h *Handlers) applyPasswordChange(
 	).First(&user).Error
 	if err != nil {
 		h.serverError(
-			w, "failed to load user for password change", err,
+			w, r, "failed to load user for password change", err,
 		)
 
 		return "", "", false
@@ -113,7 +107,7 @@ func (h *Handlers) applyPasswordChange(
 		currentPassword, user.Password,
 	)
 	if err != nil {
-		h.serverError(w, "failed to verify password", err)
+		h.serverError(w, r, "failed to verify password", err)
 
 		return "", "", false
 	}
@@ -132,7 +126,7 @@ func (h *Handlers) applyPasswordChange(
 
 	hashedPassword, err := database.HashPassword(newPassword)
 	if err != nil {
-		h.serverError(w, "failed to hash new password", err)
+		h.serverError(w, r, "failed to hash new password", err)
 
 		return "", "", false
 	}
@@ -141,7 +135,7 @@ func (h *Handlers) applyPasswordChange(
 		"password", hashedPassword,
 	).Error
 	if err != nil {
-		h.serverError(w, "failed to update password", err)
+		h.serverError(w, r, "failed to update password", err)
 
 		return "", "", false
 	}
@@ -162,7 +156,7 @@ func (h *Handlers) profileOwnerOrDeny(
 ) (string, string, bool) {
 	requestedUsername := chi.URLParam(r, "username")
 	if requestedUsername == "" {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return "", "", false
 	}
@@ -172,7 +166,7 @@ func (h *Handlers) profileOwnerOrDeny(
 	// unexpected retrieval error.
 	sess, err := h.session.Get(r)
 	if err != nil {
-		h.serverError(w, "failed to get session", err)
+		h.serverError(w, r, "failed to get session", err)
 
 		return "", "", false
 	}
@@ -180,10 +174,7 @@ func (h *Handlers) profileOwnerOrDeny(
 	sessionUsername, ok := h.session.GetUsername(sess)
 	if !ok {
 		h.log.Error("authenticated session missing username")
-		http.Error(
+		h.renderError(w, r, http.StatusInternalServerError)
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return "", "", false
 	}
@@ -191,17 +182,14 @@ func (h *Handlers) profileOwnerOrDeny(
 	sessionUserID, ok := h.session.GetUserID(sess)
 	if !ok {
 		h.log.Error("authenticated session missing user ID")
-		http.Error(
+		h.renderError(w, r, http.StatusInternalServerError)
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return "", "", false
 	}
 
 	// Only allow users to act on their own profile.
 	if requestedUsername != sessionUsername {
-		http.Error(w, "Forbidden", http.StatusForbidden)
+		h.renderError(w, r, http.StatusForbidden)
 
 		return "", "", false
 	}

internal/handlers/profile_test.go

@@ -128,7 +128,9 @@ func TestUserRoute_Unauthenticated_RedirectedByMiddleware(t *testing.T) {
 
 	var sess *session.Session
 
-	app := newTestApp(t, &log, &cfg, &sess)
+	var h *handlers.Handlers
+
+	app := newTestApp(t, &log, &cfg, &sess, &h)
 	app.RequireStart()
 
 	t.Cleanup(app.RequireStop)
@@ -139,7 +141,7 @@ func TestUserRoute_Unauthenticated_RedirectedByMiddleware(t *testing.T) {
 
 	router := chi.NewRouter()
 	router.Route("/user/{username}", func(r chi.Router) {
-		r.Use(mw.CSRF())
+		r.Use(mw.CSRF(h.HandleErrorPage(http.StatusForbidden)))
 		r.Use(mw.RequireAuth())
 		r.Get("/", func(w http.ResponseWriter, _ *http.Request) {
 			handlerReached = true

internal/handlers/source_management.go

@@ -149,13 +149,7 @@ func (h *Handlers) HandleSourceList() http.HandlerFunc {
 			"user_id = ?", userID,
 		).Order("created_at DESC").Find(&webhooks).Error
 		if err != nil {
-			h.log.Error(
+			h.serverError(w, r, "failed to list webhooks", err)
-				"failed to list webhooks", "error", err,
-			)
-			http.Error(
-				w, "Internal server error",
-				http.StatusInternalServerError,
-			)
 
 			return
 		}
@@ -249,9 +243,7 @@ func (h *Handlers) HandleSourceCreateSubmit() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err := r.ParseForm()
 		if err != nil {
-			http.Error(
+			h.renderError(w, r, http.StatusBadRequest)
-				w, "Bad request", http.StatusBadRequest,
-			)
 
 			return
 		}
@@ -311,7 +303,7 @@ func (h *Handlers) createWebhookWithEntrypoint(
 
 	err := h.commitWebhook(webhook)
 	if err != nil {
-		h.serverError(w, "failed to create webhook", err)
+		h.serverError(w, r, "failed to create webhook", err)
 
 		return
 	}
@@ -388,7 +380,7 @@ func (h *Handlers) HandleSourceDetail() http.HandlerFunc {
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -420,7 +412,7 @@ func (h *Handlers) renderSourceDetail(
 	if h.dbMgr.DBExists(webhook.ID) {
 		webhookDB, err := h.dbMgr.GetDB(webhook.ID)
 		if err != nil {
-			h.serverError(w, "failed to get webhook database", err)
+			h.serverError(w, r, "failed to get webhook database", err)
 
 			return
 		}
@@ -429,7 +421,7 @@ func (h *Handlers) renderSourceDetail(
 			webhookDB, webhook.ID, singleHTTPTargetID(targets),
 		)
 		if err != nil {
-			h.serverError(w, "failed to load recent events", err)
+			h.serverError(w, r, "failed to load recent events", err)
 
 			return
 		}
@@ -482,7 +474,7 @@ func (h *Handlers) HandleSourceEdit() http.HandlerFunc {
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -517,7 +509,7 @@ func (h *Handlers) HandleSourceEditSubmit() http.HandlerFunc {
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -526,9 +518,7 @@ func (h *Handlers) HandleSourceEditSubmit() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err = r.ParseForm()
 		if err != nil {
-			http.Error(
+			h.renderError(w, r, http.StatusBadRequest)
-				w, "Bad request", http.StatusBadRequest,
-			)
 
 			return
 		}
@@ -582,7 +572,7 @@ func (h *Handlers) applyWebhookEdit(
 
 	err := h.db.DB().Save(webhook).Error
 	if err != nil {
-		h.serverError(w, "failed to update webhook", err)
+		h.serverError(w, r, "failed to update webhook", err)
 
 		return
 	}
@@ -612,7 +602,7 @@ func (h *Handlers) HandleSourceDelete() http.HandlerFunc {
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -639,7 +629,7 @@ func (h *Handlers) deleteWebhookResources(
 	// be removed by hand; deleted history cannot be recovered.
 	err := h.commitWebhookDeletion(&webhook)
 	if err != nil {
-		h.serverError(w, "failed to delete webhook", err)
+		h.serverError(w, r, "failed to delete webhook", err)
 
 		return
 	}
@@ -665,7 +655,7 @@ func (h *Handlers) deleteWebhookResources(
 		// redirecting as though everything succeeded: the file
 		// needs removing by hand, and the logged error names it.
 		h.serverError(
-			w, "failed to delete webhook event database", err,
+			w, r, "failed to delete webhook event database", err,
 		)
 
 		return
@@ -809,7 +799,7 @@ func (h *Handlers) ownedWebhook(
 		"id = ? AND user_id = ?", sourceID, userID,
 	).First(&webhook).Error
 	if err != nil {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return database.Webhook{}, false
 	}
@@ -831,7 +821,7 @@ func (h *Handlers) HandleSourceLogs() http.HandlerFunc {
 			// Without the map every delivery renders through a
 			// zero redactor, so failing the page is the only
 			// safe answer.
-			h.serverError(w, "failed to load targets", err)
+			h.serverError(w, r, "failed to load targets", err)
 
 			return
 		}
@@ -839,7 +829,7 @@ func (h *Handlers) HandleSourceLogs() http.HandlerFunc {
 		page := h.parsePage(r)
 
 		evts, total, ok := h.loadEventsWithDeliveries(
-			w, webhook, targets, page,
+			w, r, webhook, targets, page,
 		)
 		if !ok {
 			return
@@ -949,6 +939,7 @@ func (h *Handlers) parsePage(r *http.Request) int {
 // caller must then render nothing further.
 func (h *Handlers) loadEventsWithDeliveries(
 	w http.ResponseWriter,
+	r *http.Request,
 	webhook database.Webhook,
 	targetMap map[string]eventLogTarget,
 	page int,
@@ -962,7 +953,7 @@ func (h *Handlers) loadEventsWithDeliveries(
 	webhookDB, err := h.dbMgr.GetDB(webhook.ID)
 	if err != nil {
 		h.serverError(
-			w, "failed to get webhook database", err,
+			w, r, "failed to get webhook database", err,
 		)
 
 		return nil, 0, false
@@ -999,7 +990,7 @@ func (h *Handlers) loadEventsWithDeliveries(
 	)
 	if err != nil {
 		h.serverError(
-			w, "failed to load delivery attempts", err,
+			w, r, "failed to load delivery attempts", err,
 		)
 
 		return nil, 0, false
@@ -1008,7 +999,7 @@ func (h *Handlers) loadEventsWithDeliveries(
 	resubmits, err := resubmitCounts(webhookDB, eventIDs)
 	if err != nil {
 		h.serverError(
-			w, "failed to count event resubmissions", err,
+			w, r, "failed to count event resubmissions", err,
 		)
 
 		return nil, 0, false
@@ -1231,7 +1222,7 @@ func (h *Handlers) HandleEntrypointCreate() http.HandlerFunc {
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -1240,9 +1231,7 @@ func (h *Handlers) HandleEntrypointCreate() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err = r.ParseForm()
 		if err != nil {
-			http.Error(
+			h.renderError(w, r, http.StatusBadRequest)
-				w, "Bad request", http.StatusBadRequest,
-			)
 
 			return
 		}
@@ -1258,7 +1247,7 @@ func (h *Handlers) HandleEntrypointCreate() http.HandlerFunc {
 
 		err = h.db.DB().Create(entrypoint).Error
 		if err != nil {
-			h.serverError(w, "failed to create entrypoint", err)
+			h.serverError(w, r, "failed to create entrypoint", err)
 
 			return
 		}
@@ -1289,7 +1278,7 @@ func (h *Handlers) HandleTargetCreate() http.HandlerFunc {
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -1298,9 +1287,7 @@ func (h *Handlers) HandleTargetCreate() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err = r.ParseForm()
 		if err != nil {
-			http.Error(
+			h.renderError(w, r, http.StatusBadRequest)
-				w, "Bad request", http.StatusBadRequest,
-			)
 
 			return
 		}
@@ -1371,7 +1358,7 @@ func (h *Handlers) processTargetCreate(
 
 	err = h.db.DB().Create(target).Error
 	if err != nil {
-		h.serverError(w, "failed to create target", err)
+		h.serverError(w, r, "failed to create target", err)
 
 		return
 	}
@@ -1465,7 +1452,7 @@ func (h *Handlers) buildTargetConfig(
 	case database.TargetTypeSlack:
 		return h.buildSlackTargetConfig(w, r, in.URL)
 	case database.TargetTypeDatabase:
-		return h.buildDatabaseTargetConfig(w, in.Expiry)
+		return h.buildDatabaseTargetConfig(w, r, in.Expiry)
 	case database.TargetTypeLog:
 		return "", nil
 	default:
@@ -1515,7 +1502,7 @@ func (h *Handlers) buildHTTPTargetConfig(
 		return "", err
 	}
 
-	return marshalTargetConfig(w, delivery.HTTPTargetConfig{
+	return h.marshalTargetConfig(w, r, delivery.HTTPTargetConfig{
 		URL:     in.URL,
 		Headers: headers,
 		Timeout: timeout,
@@ -1537,7 +1524,7 @@ func (h *Handlers) buildSlackTargetConfig(
 		return "", err
 	}
 
-	return marshalTargetConfig(w, delivery.SlackTargetConfig{
+	return h.marshalTargetConfig(w, r, delivery.SlackTargetConfig{
 		WebhookURL: targetURL,
 	})
 }
@@ -1591,16 +1578,14 @@ func (h *Handlers) validateTargetURL(
 
 // marshalTargetConfig serialises a target configuration for storage,
 // writing a 500 itself if it cannot.
-func marshalTargetConfig(
+func (h *Handlers) marshalTargetConfig(
 	w http.ResponseWriter,
+	r *http.Request,
 	cfg any,
 ) (string, error) {
 	configBytes, err := json.Marshal(cfg)
 	if err != nil {
-		http.Error(
+		h.serverError(w, r, "failed to encode target config", err)
-			w, "Internal server error",
-			http.StatusInternalServerError,
-		)
 
 		return "", err
 	}
@@ -1616,6 +1601,7 @@ func marshalTargetConfig(
 // expiry yields an empty config (the keep-forever default).
 func (h *Handlers) buildDatabaseTargetConfig(
 	w http.ResponseWriter,
+	r *http.Request,
 	expiry string,
 ) (string, error) {
 	expiry = strings.TrimSpace(expiry)
@@ -1634,8 +1620,8 @@ func (h *Handlers) buildDatabaseTargetConfig(
 		return "", err
 	}
 
-	return marshalTargetConfig(
+	return h.marshalTargetConfig(
-		w, map[string]any{"expiry": expiry},
+		w, r, map[string]any{"expiry": expiry},
 	)
 }
 
@@ -1689,7 +1675,7 @@ func (h *Handlers) deleteChildResource(
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
@@ -1699,11 +1685,7 @@ func (h *Handlers) deleteChildResource(
 			childID, webhook.ID,
 		).Delete(model)
 		if result.Error != nil {
-			h.log.Error(errMsg, "error", result.Error)
+			h.serverError(w, r, errMsg, result.Error)
-			http.Error(
-				w, "Internal server error",
-				http.StatusInternalServerError,
-			)
 
 			return
 		}
@@ -1793,18 +1775,14 @@ func (h *Handlers) toggleChildResource(
 			"id = ? AND user_id = ?", sourceID, userID,
 		).First(&webhook).Error
 		if err != nil {
-			http.NotFound(w, r)
+			h.renderError(w, r, http.StatusNotFound)
 
 			return
 		}
 
 		err = toggleFn(webhook.ID, childID)
 		if err != nil {
-			h.log.Error(errMsg, "error", err)
+			h.serverError(w, r, errMsg, err)
-			http.Error(
-				w, "Internal server error",
-				http.StatusInternalServerError,
-			)
 
 			return
 		}

internal/handlers/target_edit.go

@@ -88,9 +88,7 @@ func (h *Handlers) HandleTargetEditSubmit() http.HandlerFunc {
 		// middleware, which runs before CSRF parses the form.
 		err := r.ParseForm()
 		if err != nil {
-			http.Error(
+			h.renderError(w, r, http.StatusBadRequest)
-				w, "Bad request", http.StatusBadRequest,
-			)
 
 			return
 		}
@@ -157,7 +155,7 @@ func (h *Handlers) applyTargetEdit(
 
 	err = h.db.DB().Save(target).Error
 	if err != nil {
-		h.serverError(w, "failed to update target", err)
+		h.serverError(w, r, "failed to update target", err)
 
 		return
 	}
@@ -220,7 +218,7 @@ func (h *Handlers) ownedTarget(
 		chi.URLParam(r, "targetID"), webhook.ID,
 	).First(&target).Error
 	if err != nil {
-		http.NotFound(w, r)
+		h.renderError(w, r, http.StatusNotFound)
 
 		return database.Webhook{}, nil, false
 	}

internal/handlers/webhook.go

@@ -88,14 +88,14 @@ func (h *Handlers) processWebhookRequest(
 
 	headersJSON, err := json.Marshal(r.Header)
 	if err != nil {
-		h.serverError(w, "failed to serialize headers", err)
+		h.receiverError(w, "failed to serialize headers", err)
 
 		return
 	}
 
 	targets, err := h.loadActiveTargets(entrypoint.WebhookID)
 	if err != nil {
-		h.serverError(w, "failed to query targets", err)
+		h.receiverError(w, "failed to query targets", err)
 
 		return
 	}
@@ -196,7 +196,7 @@ func (h *Handlers) createAndDeliverEvent(
 		targets,
 	)
 	if err != nil {
-		h.serverError(w, "failed to store webhook event", err)
+		h.receiverError(w, "failed to store webhook event", err)
 
 		return
 	}
@@ -204,6 +204,19 @@ func (h *Handlers) createAndDeliverEvent(
 	h.finishWebhookResponse(w, event, entrypoint, tasks)
 }
 
+// receiverError logs an error and answers the sender with a plain-text
+// 500. The receiver's answers are for programs, so it never sends the
+// error page the web UI uses.
+func (h *Handlers) receiverError(
+	w http.ResponseWriter, msg string, err error,
+) {
+	h.log.Error(msg, "error", err)
+	http.Error(
+		w, "Internal server error",
+		http.StatusInternalServerError,
+	)
+}
+
 // eventSource carries the fields a new event is built from. The
 // receiver fills it from the live request; the resubmit handler fills
 // it from a stored event. Both then go through createAndFanOut, so an

internal/middleware/csrf.go

@@ -19,7 +19,7 @@ func CSRFToken(r *http.Request) string {
 // key to sign a CSRF cookie and validates a masked token submitted via
 // the "csrf_token" form field (or the "X-CSRF-Token" header) on
 // POST/PUT/PATCH/DELETE requests. Requests with an invalid or missing
-// token receive a 403 Forbidden response.
+// token are logged and answered by forbidden, which must write the 403.
 //
 // The middleware detects the client-facing transport protocol
 // per-request via reqtls.IsTLS, the single TLS predicate the session
@@ -36,7 +36,9 @@ func CSRFToken(r *http.Request) string {
 // Two gorilla/csrf instances are maintained — one with Secure cookies
 // (for TLS) and one without (for plaintext HTTP) — because the
 // csrf.Secure option is set at creation time, not per-request.
-func (m *Middleware) CSRF() func(http.Handler) http.Handler {
+func (m *Middleware) CSRF(
+	forbidden http.Handler,
+) func(http.Handler) http.Handler {
 	csrfErrorHandler := http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		// CSRF is registered ahead of RequireAuth on every route
 		// group that uses it, so this WARN is reachable by an
@@ -57,7 +59,7 @@ func (m *Middleware) CSRF() func(http.Handler) http.Handler {
 			"remote_addr", r.RemoteAddr,
 			"reason", csrf.FailureReason(r),
 		)
-		http.Error(w, "Forbidden - invalid CSRF token", http.StatusForbidden)
+		forbidden.ServeHTTP(w, r)
 	})
 
 	key := m.session.GetKey()

internal/middleware/csrf_test.go

@@ -18,6 +18,12 @@ import (
 // csrfCookieName is the gorilla/csrf cookie name.
 const csrfCookieName = "_gorilla_csrf"
 
+// forbidden stands in for the error page the server hands CSRF to
+// answer a refused request with.
+func forbidden(w http.ResponseWriter, _ *http.Request) {
+	w.WriteHeader(http.StatusForbidden)
+}
+
 // csrfGetToken performs a GET request through the CSRF middleware
 // and returns the token and cookies.
 func csrfGetToken(
@@ -98,7 +104,7 @@ func TestCSRF_GETSetsToken(t *testing.T) {
 
 	var gotToken string
 
-	handler := m.CSRF()(http.HandlerFunc(
+	handler := m.CSRF(http.HandlerFunc(forbidden))(http.HandlerFunc(
 		func(_ http.ResponseWriter, r *http.Request) {
 			gotToken = middleware.CSRFToken(r)
 		},
@@ -120,7 +126,7 @@ func TestCSRF_POSTWithValidToken(t *testing.T) {
 	t.Parallel()
 
 	m, _ := testMiddleware(t, config.EnvironmentDev)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	getReq := httptest.NewRequestWithContext(
 		context.Background(),
@@ -152,7 +158,7 @@ func csrfPOSTWithoutTokenTest(
 	t.Helper()
 
 	m, _ := testMiddleware(t, env)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	// GET to establish the CSRF cookie
 	getHandler := csrfMW(http.HandlerFunc(
@@ -209,7 +215,7 @@ func TestCSRF_POSTWithInvalidToken(t *testing.T) {
 	t.Parallel()
 
 	m, _ := testMiddleware(t, config.EnvironmentDev)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	// GET to establish the CSRF cookie
 	getHandler := csrfMW(http.HandlerFunc(
@@ -265,7 +271,7 @@ func TestCSRF_GETDoesNotValidate(t *testing.T) {
 
 	var called bool
 
-	handler := m.CSRF()(http.HandlerFunc(
+	handler := m.CSRF(http.HandlerFunc(forbidden))(http.HandlerFunc(
 		func(_ http.ResponseWriter, _ *http.Request) {
 			called = true
 		},
@@ -328,7 +334,7 @@ func csrfTookStrictPath(
 	t.Helper()
 
 	m, _ := testMiddleware(t, env)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	newReq := func(method string) *http.Request {
 		r := httptest.NewRequestWithContext(
@@ -477,7 +483,7 @@ func TestCSRF_ProdMode_PlaintextHTTP_POSTWithValidToken(
 	t.Parallel()
 
 	m, _ := testMiddleware(t, config.EnvironmentProd)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	getReq := httptest.NewRequestWithContext(
 		context.Background(),
@@ -517,7 +523,7 @@ func TestCSRF_ProdMode_BehindProxy_POSTWithValidToken(
 	t.Parallel()
 
 	m, _ := testMiddleware(t, config.EnvironmentProd)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	getReq := httptest.NewRequestWithContext(
 		context.Background(),
@@ -562,7 +568,7 @@ func TestCSRF_ProdMode_DirectTLS_POSTWithValidToken(
 	t.Parallel()
 
 	m, _ := testMiddleware(t, config.EnvironmentProd)
-	csrfMW := m.CSRF()
+	csrfMW := m.CSRF(http.HandlerFunc(forbidden))
 
 	getReq := httptest.NewRequestWithContext(
 		context.Background(),

internal/middleware/logbound_test.go

@@ -260,7 +260,9 @@ func logSites() map[string]logSite {
 			) http.Handler {
 				t.Helper()
 
-				return m.CSRF()(unreachable(t))
+				return m.CSRF(http.HandlerFunc(forbidden))(
+					unreachable(t),
+				)
 			},
 			send:       postNoToken,
 			wantStatus: http.StatusForbidden,

internal/middleware/loginguard.go

@@ -108,10 +108,10 @@ type failureWindow struct {
 //
 // A limiter that spends budget on arrival cannot protect a
 // single-admin product: behind the reverse proxy the deployment
-// requires, with TRUSTED_PROXIES unset, every client keys on the
+// requires, when TRUSTED_PROXIES does not cover it, every client
-// proxy, so a stranger trickling five POSTs a minute keeps the one
+// keys on the proxy, so a stranger trickling five POSTs a minute
-// bucket full and the operator's own correct password is answered 429
+// keeps the one bucket full and the operator's own correct password
-// forever. There is no second administrative path.
+// is answered 429 forever. There is no second administrative path.
 //
 // So budget is spent only by a FAILED verification. A correct
 // password is never throttled, whatever the counters say, which is

internal/middleware/ratelimit.go

@@ -123,9 +123,8 @@ func bucketKey(addr netip.Addr) string {
 	return prefix.String()
 }
 
-// isTrustedProxy reports whether addr belongs to a network the
+// isTrustedProxy reports whether addr belongs to a network in
-// operator listed in TRUSTED_PROXIES. The list is empty by default,
+// TRUSTED_PROXIES, which by default is the RFC 1918 private ranges.
-// so by default nothing is trusted.
 func (m *Middleware) isTrustedProxy(addr netip.Addr) bool {
 	for _, prefix := range m.params.Config.TrustedProxies {
 		if prefix.Contains(addr) {

internal/middleware/ratelimit_test.go

@@ -384,8 +384,8 @@ const (
 	// trustedProxyCIDR is the proxy network the forwarded-path
 	// tests configure, and trustedPeer an address inside it. A
 	// production deployment is required to run behind a reverse
-	// proxy with TRUSTED_PROXIES set, so this is the shape the
+	// proxy that TRUSTED_PROXIES covers, either by the default or by
-	// bucketing has to hold in.
+	// a set value, so this is the shape the bucketing has to hold in.
 	trustedProxyCIDR = "10.0.0.0/8"
 	trustedPeer      = "10.0.0.1:44444"
 )
@@ -426,8 +426,8 @@ func assertSharedBucket(
 }
 
 // TestRateLimitKey_SpoofedForwardedFromUntrustedPeer is the test
-// this gating exists for: with no trusted proxies configured (the
+// this gating exists for: from a peer that is not a trusted
-// default), a client that rotates a forwarded header on every
+// proxy, a client that rotates a forwarded header on every
 // request must stay in one bucket. If forwarded headers were
 // trusted unconditionally, each spoofed value would mint a fresh
 // bucket and the limit would stop no one.
@@ -1097,8 +1097,9 @@ func TestPostRateLimit_IPv4IndependentPerAddress(t *testing.T) {
 // that arrives from trustedPeer — a configured trusted proxy — and
 // names forwarded as its client in X-Forwarded-For. That is the
 // production path: a deployment is required to run behind a reverse
-// proxy with TRUSTED_PROXIES set, so the forwarded address, not the
+// proxy that TRUSTED_PROXIES covers, either by the default or by a
-// peer, is what the limiters bucket on there.
+// set value, so the forwarded address, not the peer, is what the
+// limiters bucket on there.
 func forwardedKeyFor(
 	t *testing.T, m *middleware.Middleware, forwarded string,
 ) string {
@@ -1178,9 +1179,9 @@ func TestRateLimitKey_ForwardedIPv6BucketsByPrefix(t *testing.T) {
 //
 // Every existing test of this fallback uses an IPv4 proxy, where
 // bucketKey is the identity function, so replacing the call with
-// peer.String() leaves the whole suite green. Only operator-listed
+// peer.String() leaves the whole suite green. Only addresses inside
-// addresses reach this line and the fallback is fail-closed, so this
+// TRUSTED_PROXIES reach this line and the fallback is fail-closed, so
-// pins behaviour rather than fixing a defect.
+// this pins behaviour rather than fixing a defect.
 func TestRateLimitKey_TrustedPeerUnusableForwardedMasksPeer(
 	t *testing.T,
 ) {

internal/middleware/recoverer.go

@@ -109,7 +109,8 @@ func (w *recoverResponseWriter) Unwrap() http.ResponseWriter {
 
 // Recoverer returns middleware that turns a handler panic into one
 // structured ERROR record and a 500, rather than a dropped
-// connection.
+// connection. The 500 is page when page is not nil, and plain text
+// when it is nil or when page panics before writing anything.
 //
 // It replaces chi's middleware.Recoverer, which does neither on a
 // current Go release. chi v1.5.5's pretty-printer scans the stack for
@@ -138,7 +139,9 @@ func (w *recoverResponseWriter) Unwrap() http.ResponseWriter {
 // set before panicking, because a request that failed must not hand
 // the client a credential; every other header is left to http.Error.
 // See https://git.eeqj.de/sneak/webhooker/issues/193.
-func (s *Middleware) Recoverer() func(http.Handler) http.Handler {
+func (s *Middleware) Recoverer(
+	page http.Handler,
+) func(http.Handler) http.Handler {
 	return func(next http.Handler) http.Handler {
 		return http.HandlerFunc(func(
 			w http.ResponseWriter,
@@ -171,6 +174,14 @@ func (s *Middleware) Recoverer() func(http.Handler) http.Handler {
 
 				rw.Header().Del("Set-Cookie")
 
+				if page != nil {
+					s.servePage(rw, r, page)
+				}
+
+				if rw.committed {
+					return
+				}
+
 				http.Error(
 					rw,
 					http.StatusText(
@@ -185,6 +196,27 @@ func (s *Middleware) Recoverer() func(http.Handler) http.Handler {
 	}
 }
 
+// servePage answers with page. A panic in page itself is logged and
+// recovered here, so the Recoverer can still send its plain 500.
+func (s *Middleware) servePage(
+	w http.ResponseWriter,
+	r *http.Request,
+	page http.Handler,
+) {
+	defer func() {
+		rvr := recover()
+		if rvr != nil {
+			s.log.Error("error page panic",
+				"panic", logfield.Truncate(
+					fmt.Sprint(rvr), maxPanicValueBytes,
+				),
+			)
+		}
+	}()
+
+	page.ServeHTTP(w, r)
+}
+
 // logPanic writes the record. Every field it can grow is truncated to
 // a fixed budget, so MaxPanicLogLineBytes holds.
 //

internal/middleware/recoverer_test.go

@@ -76,7 +76,7 @@ func newRecovererProbe(
 	// Logging outside so the recovered 500 is the status it records.
 	router.Use(chimw.RequestID)
 	router.Use(m.Logging())
-	router.Use(m.Recoverer())
+	router.Use(m.Recoverer(nil))
 	router.Get("/probe", handler)
 
 	serverErrors := new(bytes.Buffer)
@@ -637,7 +637,7 @@ func TestRecovererKeepsResponseControllerWorking(t *testing.T) {
 
 	m, _ := capturingMiddleware(t)
 
-	handler := m.Recoverer()(http.HandlerFunc(
+	handler := m.Recoverer(nil)(http.HandlerFunc(
 		func(w http.ResponseWriter, _ *http.Request) {
 			_, _ = w.Write([]byte("chunk"))
 
@@ -672,3 +672,59 @@ func TestRecovererKeepsResponseControllerWorking(t *testing.T) {
 	assert.Equal(t, http.StatusOK, resp.StatusCode)
 	assert.Equal(t, "chunk", string(body))
 }
+
+// TestRecovererAnswersWithThePage covers a recoverer given a page:
+// the panic is logged as before, and the 500 is that page.
+func TestRecovererAnswersWithThePage(t *testing.T) {
+	t.Parallel()
+
+	m, logs := capturingMiddleware(t)
+
+	page := http.HandlerFunc(
+		func(w http.ResponseWriter, _ *http.Request) {
+			w.WriteHeader(http.StatusInternalServerError)
+			_, _ = w.Write([]byte("the error page"))
+		},
+	)
+
+	w := httptest.NewRecorder()
+	m.Recoverer(page)(http.HandlerFunc(panicProbe)).ServeHTTP(
+		w, httptest.NewRequestWithContext(
+			t.Context(), http.MethodGet, "/", nil,
+		),
+	)
+
+	assert.Equal(t, http.StatusInternalServerError, w.Code)
+	assert.Equal(t, "the error page", w.Body.String())
+	assert.Contains(t, logs.String(), `"msg":"handler panic"`)
+	assert.Contains(t, logs.String(), panicMarker)
+}
+
+// TestRecovererFallsBackWhenThePagePanics covers a page that panics
+// before writing anything: both panics are logged, and the client
+// still gets the plain 500.
+func TestRecovererFallsBackWhenThePagePanics(t *testing.T) {
+	t.Parallel()
+
+	m, logs := capturingMiddleware(t)
+
+	const pagePanic = "QQERRORPAGEPANICQQ"
+
+	page := http.HandlerFunc(
+		func(http.ResponseWriter, *http.Request) {
+			panic(pagePanic)
+		},
+	)
+
+	w := httptest.NewRecorder()
+	m.Recoverer(page)(http.HandlerFunc(panicProbe)).ServeHTTP(
+		w, httptest.NewRequestWithContext(
+			t.Context(), http.MethodGet, "/", nil,
+		),
+	)
+
+	assert.Equal(t, http.StatusInternalServerError, w.Code)
+	assert.Equal(t, "Internal Server Error\n", w.Body.String())
+	assert.Contains(t, logs.String(), panicMarker)
+	assert.Contains(t, logs.String(), pagePanic)
+}

internal/server/error_page_test.go

@@ -0,0 +1,220 @@
+package server_test
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"net/url"
+	"strconv"
+	"testing"
+
+	"github.com/getsentry/sentry-go"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"sneak.berlin/go/webhooker/internal/config"
+	"sneak.berlin/go/webhooker/internal/server"
+)
+
+// The link back the error page offers: to the webhook list for a
+// signed-in user, to sign-in for anyone else.
+const (
+	backToWebhooks = `<a href="/sources" class="btn-secondary">` +
+		`Back to webhooks</a>`
+	backToSignIn = `<a href="/pages/login" class="btn-primary">` +
+		`Sign in</a>`
+)
+
+// assertErrorPage checks that w is the error page for status, in the
+// normal layout, offering link.
+func assertErrorPage(
+	t *testing.T,
+	w *httptest.ResponseRecorder,
+	status int,
+	link string,
+) {
+	t.Helper()
+
+	body := w.Body.String()
+
+	assert.Equal(t, status, w.Code)
+	assert.Equal(
+		t, "text/html; charset=utf-8", w.Header().Get("Content-Type"),
+	)
+	assert.Equal(t, "no-store", w.Header().Get("Cache-Control"))
+	assert.Contains(t, body, `<nav class="app-bar"`)
+	assert.Contains(
+		t, body, strconv.Itoa(status)+" "+http.StatusText(status),
+	)
+	assert.Contains(t, body, link)
+}
+
+func TestErrorPage_DeletedWebhook(t *testing.T) {
+	t.Parallel()
+
+	env := newTestEnv(t)
+
+	userID, _ := env.seedUser(t, "owner", "somepassword")
+	cookies := env.authCookies(t, userID, "owner")
+
+	wh := env.seedWebhook(t, userID)
+	require.NoError(t, env.db.DB().Delete(wh).Error)
+
+	w := env.get("/source/"+wh.ID, cookies)
+
+	assertErrorPage(t, w, http.StatusNotFound, backToWebhooks)
+}
+
+func TestErrorPage_DeletedTarget(t *testing.T) {
+	t.Parallel()
+
+	env := newTestEnv(t)
+
+	userID, _ := env.seedUser(t, "owner", "somepassword")
+	cookies := env.authCookies(t, userID, "owner")
+
+	wh := env.seedWebhook(t, userID)
+	tgt := env.seedTarget(t, wh.ID)
+	require.NoError(t, env.db.DB().Delete(tgt).Error)
+
+	w := env.get(
+		"/source/"+wh.ID+"/targets/"+tgt.ID+"/edit", cookies,
+	)
+
+	assertErrorPage(t, w, http.StatusNotFound, backToWebhooks)
+}
+
+func TestErrorPage_UnknownPath(t *testing.T) {
+	t.Parallel()
+
+	env := newTestEnv(t)
+
+	userID, _ := env.seedUser(t, "owner", "somepassword")
+	cookies := env.authCookies(t, userID, "owner")
+
+	assertErrorPage(
+		t, env.get("/no-such-page", nil),
+		http.StatusNotFound, backToSignIn,
+	)
+
+	// Outside every route group there is no form token, so the
+	// page leaves out the logout form rather than offer one that
+	// would be refused.
+	w := env.get("/no-such-page", cookies)
+	assertErrorPage(t, w, http.StatusNotFound, backToWebhooks)
+	assert.NotContains(t, w.Body.String(), `action="/pages/logout"`)
+
+	// Inside a route group the page has a token, and logout works.
+	wh := env.seedWebhook(t, userID)
+	w = env.get("/source/"+wh.ID+"/no-such-page", cookies)
+	assertErrorPage(t, w, http.StatusNotFound, backToWebhooks)
+	assert.Contains(t, w.Body.String(), `action="/pages/logout"`)
+}
+
+func TestErrorPage_BadCSRFToken(t *testing.T) {
+	t.Parallel()
+
+	env := newTestEnv(t)
+
+	form := url.Values{}
+	form.Set("username", "someone")
+	form.Set("password", "irrelevant")
+	form.Set("csrf_token", "not-a-token")
+
+	assertErrorPage(
+		t, env.post("/pages/login", form, nil),
+		http.StatusForbidden, backToSignIn,
+	)
+
+	userID, _ := env.seedUser(t, "owner", "somepassword")
+	cookies := env.authCookies(t, userID, "owner")
+	wh := env.seedWebhook(t, userID)
+
+	edit := url.Values{}
+	edit.Set("name", "renamed")
+
+	assertErrorPage(
+		t, env.post("/source/"+wh.ID+"/edit", edit, cookies),
+		http.StatusForbidden, backToWebhooks,
+	)
+}
+
+// TestErrorPage_PanicOnAdminPage sends a panicking handler in an
+// admin page route group through the real router, with error
+// tracking on: the client gets the 500 error page, and the tracker
+// still gets the panic, once. The same panic outside the admin page
+// route groups keeps the plain 500.
+func TestErrorPage_PanicOnAdminPage(t *testing.T) {
+	t.Parallel()
+
+	env := newTestEnv(t)
+
+	transport := &captureTransport{}
+
+	opts := server.SentryClientOptionsForTest(
+		"https://public@sentry.invalid/1", "webhooker-test",
+	)
+	opts.Transport = transport
+
+	client, err := sentry.NewClient(opts)
+	require.NoError(t, err)
+
+	serve := func(router http.Handler, path string) *httptest.ResponseRecorder {
+		req := httptest.NewRequestWithContext(
+			sentry.SetHubOnContext(
+				context.Background(),
+				sentry.NewHub(client, sentry.NewScope()),
+			),
+			http.MethodGet, path, nil,
+		)
+
+		w := httptest.NewRecorder()
+		router.ServeHTTP(w, req)
+
+		return w
+	}
+
+	w := serve(
+		server.NewRouterWithPageProbeForTest(
+			env.log.Get(), env.cfg, env.mw, env.hnd,
+			true, panicProbeHandler,
+		),
+		server.PageProbePattern,
+	)
+	assertErrorPage(t, w, http.StatusInternalServerError, backToSignIn)
+
+	w = serve(
+		server.NewRouterWithProbeForTest(
+			env.log.Get(), env.cfg, env.mw, env.hnd,
+			true, panicProbeHandler,
+		),
+		server.ProbePattern,
+	)
+	assert.Equal(t, http.StatusInternalServerError, w.Code)
+	assert.Equal(t, "Internal Server Error\n", w.Body.String())
+
+	require.Len(t, transport.events, 2)
+
+	for _, event := range transport.events {
+		assert.Contains(t, marshalEvent(t, event), panicProbeMarker)
+	}
+}
+
+// TestErrorPage_ReceiverStaysPlain pins that the error page is for
+// the web UI only: a sender posting to an entrypoint that does not
+// exist still gets the plain-text answer.
+func TestErrorPage_ReceiverStaysPlain(t *testing.T) {
+	t.Parallel()
+
+	// newTestEnv leaves the receiver rate limit at zero, which
+	// refuses every request before it reaches the receiver.
+	env := newTestEnvWithConfig(t, &config.Config{
+		DataDir:           t.TempDir(),
+		Environment:       config.EnvironmentDev,
+		ReceiverRateLimit: 10,
+	})
+
+	w := env.post("/webhook/no-such-entrypoint", url.Values{}, nil)
+
+	assert.Equal(t, http.StatusNotFound, w.Code)
+	assert.Equal(t, "404 page not found\n", w.Body.String())
+}

internal/server/export_test.go

@@ -5,6 +5,7 @@ import (
 	"net/http"
 
 	"github.com/getsentry/sentry-go"
+	"github.com/go-chi/chi"
 	"sneak.berlin/go/webhooker/internal/config"
 	"sneak.berlin/go/webhooker/internal/handlers"
 	"sneak.berlin/go/webhooker/internal/middleware"
@@ -101,3 +102,39 @@ func NewRouterWithProbeForTest(
 
 	return s.router
 }
+
+// PageProbePattern is where NewRouterWithPageProbeForTest serves its
+// probe: inside the /pages route group, the admin page group a
+// request reaches without signing in.
+const PageProbePattern = "/pages/probe"
+
+// NewRouterWithPageProbeForTest is NewRouterWithProbeForTest with the
+// probe added to the /pages route group once SetupRoutes has built
+// it, so the probe runs behind that group's own middleware exactly as
+// the group's real routes do.
+func NewRouterWithPageProbeForTest(
+	log *slog.Logger,
+	cfg *config.Config,
+	mw *middleware.Middleware,
+	h *handlers.Handlers,
+	sentryEnabled bool,
+	probe http.HandlerFunc,
+) http.Handler {
+	s := &Server{
+		log:    log,
+		mw:     mw,
+		h:      h,
+		params: ServerParams{Config: cfg},
+	}
+	s.sentryEnabled.Store(sentryEnabled)
+	s.SetupRoutes()
+
+	for _, route := range s.router.Routes() {
+		pages, ok := route.SubRoutes.(chi.Router)
+		if ok && route.Pattern == "/pages/*" {
+			pages.Get("/probe", probe)
+		}
+	}
+
+	return s.router
+}

internal/server/routes.go

@@ -16,8 +16,8 @@ import (
 // submission while preventing abuse from oversized payloads.
 //
 // Every route group below installs MaxBodySize(maxFormBodySize) as
-// its FIRST middleware, ahead of both CSRF and RequireAuth. Both
+// its first middleware after the recoverer, ahead of both CSRF and
-// orderings are deliberate.
+// RequireAuth. Both orderings are deliberate.
 //
 // Ahead of CSRF because gorilla/csrf parses the form. The cap has to
 // be installed before anything reads the body, or the parse runs
@@ -46,6 +46,14 @@ const requestTimeout = 60 * time.Second
 // server's router.
 func (s *Server) SetupRoutes() {
 	s.router = chi.NewRouter()
+
+	// An unknown path gets the error page. Registered before the
+	// global middleware, because chi wraps a not-found handler in the
+	// middleware already on its router, which would then run twice.
+	// The route groups below wrap it in their own middleware the same
+	// way; running theirs twice is harmless.
+	s.router.NotFound(s.h.HandleErrorPage(http.StatusNotFound))
+
 	s.setupGlobalMiddleware()
 	s.setupRoutes()
 }
@@ -69,23 +77,33 @@ func (s *Server) setupGlobalMiddleware() {
 	// Panic recovery, deliberately here rather than first. It has to
 	// run inside every middleware that observes the response, so the
 	// 500 it writes is the status the access log records and the
-	// metrics count, and outside the sentryhttp handler below, whose
+	// metrics count, and outside the sentryhttp handler, whose
 	// Repanic option needs something further out to catch what it
 	// re-raises. chi's own middleware.Recoverer held the first slot
 	// until it was measured: on a current Go release it crashes
 	// inside its stack pretty-printer instead of recovering, so the
 	// connection dropped and the original panic was never reported.
 	// See https://git.eeqj.de/sneak/webhooker/issues/187.
-	s.router.Use(s.mw.Recoverer())
+	s.recoverPanics(s.router, nil)
+}
+
+// recoverPanics installs on r the recoverer, answering a panic with
+// page (a plain 500 when page is nil), and inside it the Sentry error
+// reporting (if SENTRY_DSN is set). Repanic is true so panics still
+// bubble up to the recoverer.
+//
+// Each admin page route group installs its own, with the error page,
+// as its first middleware. A panic there is logged, reported and
+// answered inside the group and never reaches the global recoverer,
+// which keeps the plain 500 for every other route.
+func (s *Server) recoverPanics(r chi.Router, page http.Handler) {
+	r.Use(s.mw.Recoverer(page))
 
-	// Sentry error reporting (if SENTRY_DSN is set). Repanic is
-	// true so panics still bubble up to the Recoverer middleware
-	// registered immediately above.
 	if s.sentryEnabled.Load() {
 		sentryHandler := sentryhttp.New(sentryhttp.Options{
 			Repanic: true,
 		})
-		s.router.Use(sentryHandler.Handle)
+		r.Use(sentryHandler.Handle)
 	}
 }
 
@@ -147,18 +165,22 @@ func (s *Server) setupRoutes() {
 
 func (s *Server) setupPageRoutes() {
 	s.router.Route("/pages", func(r chi.Router) {
+		s.recoverPanics(
+			r, s.h.HandleErrorPage(http.StatusInternalServerError),
+		)
 		// MaxBodySize precedes CSRF and RequireAuth deliberately;
 		// see maxFormBodySize for why, and for what it costs.
 		r.Use(s.mw.MaxBodySize(maxFormBodySize))
-		r.Use(s.mw.CSRF())
+		r.Use(s.mw.CSRF(s.h.HandleErrorPage(http.StatusForbidden)))
 		r.Use(s.mw.NoCache())
 
 		// The login POST carries no pre-emptive rate limiter. Behind
-		// the reverse proxy production requires, with TRUSTED_PROXIES
+		// the reverse proxy production requires, when TRUSTED_PROXIES
-		// unset, every client shares one bucket, so a limiter spent
+		// does not cover it, every client shares one bucket, so a
-		// on arrival lets any stranger deny the operator the only
+		// limiter spent on arrival lets any stranger deny the operator
-		// administrative path. The handler verifies credentials first
+		// the only administrative path. The handler verifies
-		// and charges only failures; see Handlers.authenticateUser.
+		// credentials first and charges only failures; see
+		// Handlers.authenticateUser.
 		r.Get("/login", s.h.HandleLoginPage())
 		r.Post("/login", s.h.HandleLoginSubmit())
 
@@ -168,10 +190,13 @@ func (s *Server) setupPageRoutes() {
 
 func (s *Server) setupUserRoutes() {
 	s.router.Route("/user/{username}", func(r chi.Router) {
+		s.recoverPanics(
+			r, s.h.HandleErrorPage(http.StatusInternalServerError),
+		)
 		// MaxBodySize precedes CSRF and RequireAuth deliberately;
 		// see maxFormBodySize for why, and for what it costs.
 		r.Use(s.mw.MaxBodySize(maxFormBodySize))
-		r.Use(s.mw.CSRF())
+		r.Use(s.mw.CSRF(s.h.HandleErrorPage(http.StatusForbidden)))
 		r.Use(s.mw.NoCache())
 		r.Use(s.mw.RequireAuth())
 		r.Get("/", s.h.HandleProfile())
@@ -183,10 +208,13 @@ func (s *Server) setupUserRoutes() {
 
 func (s *Server) setupSourceRoutes() {
 	s.router.Route("/sources", func(r chi.Router) {
+		s.recoverPanics(
+			r, s.h.HandleErrorPage(http.StatusInternalServerError),
+		)
 		// MaxBodySize precedes CSRF and RequireAuth deliberately;
 		// see maxFormBodySize for why, and for what it costs.
 		r.Use(s.mw.MaxBodySize(maxFormBodySize))
-		r.Use(s.mw.CSRF())
+		r.Use(s.mw.CSRF(s.h.HandleErrorPage(http.StatusForbidden)))
 		r.Use(s.mw.NoCache())
 		r.Use(s.mw.RequireAuth())
 		r.Get("/", s.h.HandleSourceList())
@@ -195,10 +223,13 @@ func (s *Server) setupSourceRoutes() {
 	})
 
 	s.router.Route("/source/{sourceID}", func(r chi.Router) {
+		s.recoverPanics(
+			r, s.h.HandleErrorPage(http.StatusInternalServerError),
+		)
 		// MaxBodySize precedes CSRF and RequireAuth deliberately;
 		// see maxFormBodySize for why, and for what it costs.
 		r.Use(s.mw.MaxBodySize(maxFormBodySize))
-		r.Use(s.mw.CSRF())
+		r.Use(s.mw.CSRF(s.h.HandleErrorPage(http.StatusForbidden)))
 		r.Use(s.mw.NoCache())
 		r.Use(s.mw.RequireAuth())
 		r.Get("/", s.h.HandleSourceDetail())

templates/error.html

@@ -0,0 +1,15 @@
+{{template "base" .}}
+
+{{define "title"}}{{.StatusText}} - Webhooker{{end}}
+
+{{define "content"}}
+<div class="max-w-4xl mx-auto px-6 py-12">
+    <h1 class="text-2xl font-medium text-gray-900 mb-4">{{.Status}} {{.StatusText}}</h1>
+    <p class="text-gray-600 mb-6">{{.Message}}</p>
+    {{if .User}}
+    <a href="/sources" class="btn-secondary">Back to webhooks</a>
+    {{else}}
+    <a href="/pages/login" class="btn-primary">Sign in</a>
+    {{end}}
+</div>
+{{end}}

templates/navbar.html

@@ -24,10 +24,14 @@
                 </svg>
                 {{.User.Username}}
             </a>
+            {{/* An error page can be served before a form token is issued,
+                 and a logout without one is refused. */}}
+            {{if .CSRFToken}}
             <form method="POST" action="/pages/logout" class="inline">
                 <input type="hidden" name="csrf_token" value="{{.CSRFToken}}">
                 <button type="submit" class="btn-text">Logout</button>
             </form>
+            {{end}}
             {{else}}
             <a href="/pages/login" class="btn-primary">Login</a>
             {{end}}
@@ -40,10 +44,12 @@
             {{if .User}}
             <a href="/sources" class="btn-text w-full text-left">Webhooks</a>
             <a href="/user/{{.User.Username}}" class="btn-text w-full text-left">Profile</a>
+            {{if .CSRFToken}}
             <form method="POST" action="/pages/logout">
                 <input type="hidden" name="csrf_token" value="{{.CSRFToken}}">
                 <button type="submit" class="btn-text w-full text-left">Logout</button>
             </form>
+            {{end}}
             {{else}}
             <a href="/pages/login" class="btn-primary w-full">Login</a>
             {{end}}