Unbounded client-controlled values reached the 2000-entry admin log ring
buffer and its SSE fan-out, letting an unauthenticated burst pin large amounts
of heap. A boundRequestID middleware now drops an inbound X-Request-Id over
128 bytes or outside printable ASCII, so chi generates its own, and the logged
request path is capped at 256 bytes. Both hunks are needed: a raw-socket probe
showed a 1MB r.URL.Path reaches the same sink independently of the header.
Verified by a panel of agents; an unpatched-tree reproduction fails 3 of the 4
added tests with the attacker bytes visible in the log record. UUID, 32-hex,
W3C traceparent and chi's own generated id format all still pass unchanged.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* feat(auth): add revocable API tokens (bot/service auth)
Add long-lived, revocable API tokens so headless clients (the introspection
MCP tool, bots, CI) can authenticate without a password. Presented as
"Authorization: Bearer <token>", a token authenticates as a specific user,
inheriting that user's role and permissions.
- migration 018 + dedicated api_tokens table (kept separate from sessions so
bulk logout and the per-user session cap never touch these); only the
SHA-256 hash is stored, raw token shown once at creation
- auth.ResolveTokenHash: one shared bearer resolver that both AuthMiddleware
and adminAuthMiddleware now call. Sessions are matched first so existing
login behavior is unchanged; API tokens are a fallback only on session miss.
A DB outage is returned wrapped, never mistaken for a bad token.
- `server token create|list|revoke` CLI: mints directly against the DB with no
HTTP and no login — the password-free bootstrap path
- tests: resolver (8 cases incl. outage-not-fallthrough), db queries (6),
api middleware integration (valid + revoked token)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(tools): add owncord-introspect MCP server
A local MCP dev tool that lets Claude Code introspect a running OwnCord
instance: read its logs, query any REST endpoint, and tail the desktop
client's log file. It is a thin wrapper over the existing API plus the
client log — no new product surface.
- tools/mcp-introspect/index.mjs (Node/ESM, one dep: @modelcontextprotocol/sdk)
exposes api_request (full read-write passthrough), server_logs (admin SSE
ring-buffer stream), client_logs (reads the desktop log file)
- authenticates with an API token (OWNCORD_API_TOKEN); pins the self-signed
cert and skips hostname checks (the cert has no SAN)
- registered in .mcp.json (secret-free ${OWNCORD_API_TOKEN})
- un-ignore tools/mcp-introspect/ so this shared dev tool is committed, while
tools/livekit-server.exe and node_modules stay ignored
- docs/mcp-introspect.md: how it works, tool reference, setup, troubleshooting
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(dependencies): update and add various crate versions in Cargo.lock
* feat(admin): manage API tokens from the admin panel
Add Owner-gated HTTP endpoints and a UI card to create, list, and revoke
API tokens from the web admin panel. Previously only the `server token`
CLI could manage them, which requires shell access to the host.
- POST|GET|DELETE /admin/api/tokens in admin/handlers_tokens.go, wired in
admin/api.go. All three are Owner-only (ownerOnlyMiddleware, like
backups/updates): an HTTP token-mint endpoint is a network-reachable
credential-minting surface, and API tokens deliberately survive password
change + bulk logout, so a hijacked admin session must not mint one.
- Reuses the same db.*APIToken calls as the CLI; create sources the actor
from request context (audits who clicked, not the bound user); the raw
token is returned once in the 201 body, never stored.
- Add json tags to db.APITokenListItem for snake_case wire consistency.
- Admin panel: "API Tokens" nav item + create modal, show-once reveal,
revoke confirm in admin/static/index.html.
- Tests: 7 in admin/api_test.go (+api_tokens table in the in-memory schema).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* refactor: modernize to Go 1.26 idioms + enable modernize linter
Apply `golangci-lint modernize` autofixes across the server and enable the
linter in .golangci.yml so these stop re-accumulating (they built up only
because modernize was never in the config).
Production code: slices.Contains for hand-rolled membership loops (api
router, ws origin, db/account, plugin manifest); strings.SplitSeq for
allocation-free line/segment iteration (db/migrate, updater, livekit_proxy);
strings.Cut (config); fmt.Appendf (dm_handler); min() (event_pruner);
any (ws client). Tests: range-over-int, t.Context(), WaitGroup.Go,
slices.Sort, maps.Copy, new(expr), interface{}->any.
- plugin/manifest.go parent-traversal check applied by hand: modernize
skipped it (two conflicting rewrites); used the slices.Contains form.
- Removed the now-dead ptr() test helper after newexpr inlined its callers.
- Dropped dangling sort imports left by the sort.Slice->slices.Sort rewrite.
No behavior change. All four tag variants build, full test suite is green,
and golangci-lint (with modernize enabled) reports 0 issues.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(admin): reject banned users in admin auth (F1)
adminAuthMiddleware accepted a Bearer token on session validity plus the
ADMINISTRATOR bit alone and never consulted ban state, so a ban never
revoked admin-panel access. Adds the auth.IsEffectivelyBanned guard that
api.AuthMiddleware already uses, at both admin credential-resolution points.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(ws): gate the voice-channel text subscription on READ_MESSAGES (F2)
registerNow subscribed any client with voice state to that channel's
text-message topic regardless of READ_MESSAGES. The handshake's
already-computed readable-channel set is now passed into registerNow and the
subscription only happens when the voice channel is in it, preserving
authorized reconnect delivery.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(service): require READ_MESSAGES to delete messages (F4)
The non-DM delete gate checked MANAGE_MESSAGES without READ_MESSAGES, so a
role locked out of a private channel could still delete every message in it.
Requires ReadMessages alongside ManageMessages (and alongside SendMessages on
the author path) and derives the mod flag from that same gate.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(service): require READ_MESSAGES alongside MANAGE_MESSAGES in SetMessagePinned (F8)
Pin/unpin checked only MANAGE_MESSAGES, so a role denied READ on a private
channel could still pin and unpin its messages.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(service): enforce the DM block at every DM interaction sink (F5)
The DM block was only checked on send, leaving edit, reactions, pins and
typing as bypasses. One shared requireDMNotBlocked is now called from all of
them.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(ws): re-check CONNECT_VOICE when minting a refreshed LiveKit token (F6)
voice_token_refresh re-minted a LiveKit token without re-checking
CONNECT_VOICE, so a revoked permission kept working for the life of the
session. The permission is now re-checked where the token is minted, and a
60s sweep evicts participants whose permission was revoked.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(ws): rate-limit voice_e2ee_offer after validation, keyed on server state (F7)
The limiter key was built from unvalidated client input, letting an attacker
grow the limiter map without bound. The limiter now runs after validation and
keys on (sender, voiceChannelID), never on client input.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(ws): deliver voice_state/voice_leave only to roles that may read the channel (F9)
Voice state of private channels was broadcast to every connected client,
leaking channel membership. All 11 emit sites now route through one
READ-filtered fan-out, channel-tagged so replay filters too.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(api): redact the LiveKit access token from proxy dial-failure logs (F10)
A dial failure wrote the LiveKit access-token JWT into the server log via the
URL in the error. redactKey now runs on the error before it reaches slog.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(auth): reserve the [deleted-N] username namespace (F11, F12)
The tombstone username namespace used by account deletion was freely
registrable, letting a user impersonate a deleted account. The namespace is
now reserved at validation, and DeleteAccount retries with a random suffix on
collision.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(api): strip Unicode format characters from upload filenames (F13)
The attachment filename sanitizer stripped control characters but not
unicode.Cf, allowing bidi-override extension spoofing. Cf is now stripped
alongside controls and foreign path separators are cut.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* fix(api): reserve the login attempt before the bcrypt compare (F3)
The per-username lockout was a read-only IsLockedOut check followed by a
failure recorded only after the ~250ms bcrypt compare, so N concurrent
requests all passed the stale check before any of them recorded a failure.
The per-username cap is the only cross-IP brute-force defence (the middleware
limits per IP), so a distributed burst landed N guesses per 15-minute window
instead of 10.
Both counters are now reserved atomically with limiter.Allow before the
compare, and the lockout decision moves to the read-only limiter.Check so the
reservation is not double-counted. The limits are sized at threshold+1, which
leaves the sequential accepted-input set byte-identical to the previous
behaviour: failures 1-10 still land, the 10th still trips the lockout, and the
account owner's correct password on attempt 10 still returns 200. Sizing at
threshold instead would make 9 cheap wrong guesses convert the victim's own
correct password into a 15-minute lockout - the regression that got two
earlier attempts at this fix rejected, now pinned by a boundary test.
Deliberately scoped to handleLogin. The report also suggested widening to the
password-confirmation endpoints, but those are authenticated, share a single
pw_confirm_fail key across the TOTP endpoints, and widening there is what got
the first attempt rejected.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* chore(deps): bump five Rust dependencies in /Client/tauri-client/src-tauri
Rolls up dependabot #1259, #1260, #1261, #1262 and #1263:
tauri-build 2.5.6 -> 2.6.3
tauri-plugin-fs 2.4.5 -> 2.5.1
tauri-plugin-http 2.5.7 -> 2.5.9
tauri-plugin-store 2.4.2 -> 2.4.4
webpki-roots 1.0.6 -> 1.0.9
All five are lockfile-only; the manifest constraints already permitted the
new versions. The five PRs each rewrote overlapping regions of the same
Cargo.lock and so could not be merged independently, so the lockfile was
regenerated with cargo update --precise for each crate instead. The combined
result is smaller than the sum of the five diffs because they share
transitive updates.
Verified with cargo check --locked --all-targets (exit 0).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* chore(deps): bump typescript-eslint from 8.58.0 to 8.65.0 in /Client/tauri-client
Dependabot #1258.
8.65.0 improves @typescript-eslint/no-unnecessary-type-assertion, which
surfaces four assertions that were already redundant and now fail the lint
gate. They are removed here rather than in a follow-up so no commit in this
branch leaves `npm run lint` red:
UserBar.ts / members.store.ts "online" as UserStatus -> "online"
(the receiver already accepts the literal)
media.ts drops `as RequestInit` on a literal that is
already assignable
LoginForm.ts drops `as { message: unknown }` made
redundant by the `"message" in err` narrowing
All four are the rule's own autofix. Verified: npm run typecheck, npm run
lint, npm run format:check all clean, and the unit suite is 3572/3572 green
across 129 files.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
A pre-merge review caught things my local verification missed, because two CI
gates could not run in the sandbox and I mis-read a third.
golangci-lint (BLOCKER — would have turned CI red on both server matrix legs).
My local binary was built for Go 1.25 against a repo targeting 1.26, so I could
not run it. Installed 2.11.3 with the repo's own toolchain: 5 issues, all in
files this PR adds, base clean. Now 0 issues:
- bodyclose x3 in api/livekit_proxy_ws_test.go — websocket.Dial's *http.Response
was discarded; adopt the repo's existing pattern from ws/ws_integration_test.go
- gocritic stringXbytes — string(got) != string(payload) -> !bytes.Equal
- staticcheck SA4000 in ws/topic_rate_limiter_test.go — `!Allow() || !Allow()`.
This was a real defect, not just a lint: || short-circuits, so a failing first
call skipped the second, left a token unspent, and the next assertion would
have reported the wrong thing. Split into two statements.
Playwright: I reported this suite as passing. It does not. I read the exit code
of `tail` through a pipeline instead of playwright's own. Re-run properly: 229
of 255 web tests fail, all cascading from the shared login helper
(navigateToMainPage never sees [data-testid='app-layout']). It reproduces on a
clean b3caceb worktree, so it is pre-existing on main and unrelated to this
diff — but it was never true that I had verified it. Recorded as new finding
T-2026-07-25-21 and promoted to backlog #2; the client-e2e job stays
continue-on-error and now carries timeout-minutes so a red suite cannot burn
unbounded Actions minutes. rust-tests gets a timeout too.
Audit-doc corrections (all confirmed by re-measurement):
- screenShare.ts was listed as "untouched by this pass" at 61.1% when this PR
takes it to 100%; T-12's wording made it the exception when it is the best
- IsEitherBlocked was NOT zero-coverage — 83.3% at base via message_test.go
- excluded LOC 2,229 -> 1,827
- "40 Playwright spec files" -> 44 (33 web + 11 native), 255 web tests
- HandleLiveKitHealthForTest callers: eight -> seven
- admin coverage: 71.4% is with only the T-01 fix; 77.9% with this PR's tests
Verified after the fixes: golangci-lint 0 issues, go vet, go test -race,
go test -tags deadlock, vitest 94.87%, cargo test --lib 74/74, tsc, prettier.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AEETs3Vh6sAHHb1jMBL75g
Audits what actually has tests, then closes the gaps it found. Full write-up
with before/after numbers in docs/audit-test-coverage-2026-07-25.md.
Measurement first: `go test ./... -coverprofile` (what CI runs) instruments
each package only for itself, so code exercised through another package's
tests reads as uncovered — `service` reported 36.7% against a real 85%. All
analysis here uses -coverpkg=./..., and both views now have Makefile targets.
Features that had zero coverage at every layer:
- user blocking (db + service + the /api/v1/blocks routes)
- auth lockout persistence — the DB round-trip that survives a restart
- plugin install/enable/disable/uninstall and the plugin KV namespace
- event replay bounds (GetMaxEventSeq, PruneEventsOlderThan)
- LiveKit participant_joined webhook (replayed-token guard), the room-service
client, and proxyWebSocket/copyWS
- ws_proxy.rs and livekit_proxy.rs — pure helpers extracted, matching the
existing tofu.rs pattern, so cert-pin and header-injection checks are testable
Gaps that were hidden rather than absent:
- Server/admin reported 0.3% coverage with 307 tests passing. TestSpawnDetached_*
re-execs the test binary; the child inherited GOCOVERDIR and the parent's
stdout, clobbering the profile and printing "[no tests to run]". Now 71.4%,
and CI's uploaded artifact is correct.
- vitest.config.ts excluded 2.2k LOC unexplained, including two files that
already had tests. Trimmed to three entries, each justified inline.
- api.HandleLiveKitHealthForTest re-implemented the handler it claimed to
expose, so eight call sites tested a copy. Added a hook to the real one.
Two bugs found and pinned rather than silently patched: logctx.WithGroup nests
req_id under the group, and drag-reorder.ts takes one listener ref per channel
but releases one per sidebar, so the count never reaches zero.
Coverage: client 92.93% -> 94.87% statements (3371 -> 3572 tests) even after
un-excluding hidden files; Rust 47 -> 74 tests; Go zero-coverage functions
~70 -> 21, with plugin 61->77%, admin 67->86%, db 76->84%, service 85->91%.
Verified: go vet, all four build-tag variants, go test -race, -tags deadlock,
vitest --coverage, cargo test --lib, cargo clippy --all-targets, playwright.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AEETs3Vh6sAHHb1jMBL75g
Rebasing onto post-F3 main surfaced two spots the auto-merge left inconsistent:
- profile_handler UpdateIdentityKey path: thread ctx into the writeServiceError
call (the signature gained a context param in the server logging change)
- identity-pin store lookup: drop a needless borrow flagged by clippy
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make server failures debuggable without leaking secrets:
- configurable stdout log level (config.yaml logging.level + OWNCORD_LOGGING_LEVEL)
- preserve the DB cause in ErrInternal wraps; log auth-DB failures distinctly
from bad tokens; log the previously-silent expired-session cleanup goroutine
- route HTTP handler panics through slog (was chi stderr-only, invisible to
the admin log stream)
- stackutil: argument-free panic stacks so key/token bytes never reach the
admin ring buffer / SSE; slog.LogValuer redaction on VoiceConfig/GitHubConfig/
GIFConfig/Config and db.User/db.Session
- logctx: req_id/trace_id correlation on ...Context log calls
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- client: update endpoint now sends {{target}}-{{arch}}-{{bundle_type}} so the
server-echoed platforms key matches the updater plugin's
{os}-{arch}-{installer} lookup (previously bare {{target}} produced a key
the plugin never matches, so no update was ever surfaced)
- client: TOFU cert pin is scoped to the OwnCord server host via
HostScopedVerifier; the GitHub installer download validates against web PKI
instead of failing the pinned-fingerprint check on every install
- client: check/install share one build_updater helper so the two paths cannot
diverge; tauri-plugin-updater minor-pinned per its configure_client guidance
- server: client-update endpoint serves target-specific artifacts (NSIS,
per-arch AppImage) and returns 204 for targets without a published updater
artifact (deb, darwin) instead of always serving the Windows NSIS installer
- release: server-update-manifest.json now binds both OS assets (legacy
top-level pair kept pointing at the Windows binary so deployed servers still
verify); VerifyReleaseManifest resolves the entry matching the downloaded
asset, fixing Linux server self-update
- release: ARM64 staging renames installer, tar.gz and .sig consistently so
signatures keep pairing and arch-less names cannot collide with x86_64 assets
- ci: run cargo test --lib (Rust #[cfg(test)] code was never compiled in CI);
merge the two ptt tests that raced on the global PTT_VKEY atomic
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes all 109 golangci-lint findings (106 contextcheck, 1 gocritic,
2 gosec) that accumulated after D2 wired dbgen (whose queries take ctx)
under ctx-less db.DB wrappers while CI lint was quota-dead. No nolint
comments added; every finding fixed by genuinely threading context.
- db: all 138 hand-written db.DB methods take ctx first; the dbCtx()
Background shim is deleted; raw Query/QueryRow/Exec/Begin use their
Context variants; the four redundant ctx-less passthroughs removed.
db.Auditor/WriteAudit gain ctx.
- Seams: permissions.Checker (DB iface, HasChannelPerm,
RequireChannelAccess) and the service.Store interface mirror the new
signatures (ws.EventStore and plugin.PluginStore already did).
- Callers: api/admin handlers use r.Context(); ws per-message paths use
the connection ctx via DispatchV2; hub loops and startup wiring use
context.Background(); service methods thread ctx where they have one
and Background where no ctx exists. Public service surface reached by
ctx-holding chains (PermissionService.HasChannelPerm/GetRoleForUser/
RequireChannelAccess, message/dm/block/invite/profile methods) is now
ctx-first.
- Detached (context.WithoutCancel) where cancellation would break an
invariant, found by a 3-lens adversarial review of the diff:
* voice-leave background retries (a dead webhook/connection ctx killed
retry 2 before it ran, leaving ghost capacity-holding voice rows)
* rollbackVoiceJoin's compensating delete (its trigger IS the cancel)
* post-2FA-change DeleteOtherSessions and logout DeleteSession (the
security tail of a committed change must not die with the request)
* all api/ws audit writes (a banned user could suppress their own
login_blocked_banned row by aborting the request mid-bcrypt)
* admin backup VACUUM INTO (an interrupt left a truncated .db that
the backup list presented as restorable)
* post-commit message/edit refetches (a committed message must still
fan out when the sender disconnects)
* hub settings-cache refresh (one dead connection could pin stale
values for the 30s TTL)
- gocritic rangeValCopy fixed (index iteration); gosec G306 excluded in
config with justification (generated source must stay world-readable)
instead of flipping genprotocol output to 0o600.
Verified: gofmt/vet, all four build-tag variants, full suite, deadlock
pass, full -race pass, golangci-lint 0 issues uncapped.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Applied from a deadcode (RTA from mains, all build tags) sweep with
per-symbol adversarial verification:
Deleted (nothing but their own self-tests used them):
- admin.Handler (deprecated since Phase 6; production mounts NewHandler)
plus its two self-tests
- ws.Hub.broadcastVoiceStateUpdate + wrapper + two self-tests (pre-V2
leftover; the live voice_state path is the hub voice routines)
- ws.VoiceLeaveEvent + methods ('retained as scaffolding', never
constructed in production; MsgTypeVoiceLeaveBC stays — live via the
leave routine)
- ws.parseIdentity (production calls parseParticipantIdentity directly;
ParseIdentityForTest now exercises the real parser)
- telemetry.Float64 (String/Int64 are used; the float case is covered by
the otel-tagged internal test, re-addable when a caller appears)
Moved into export_test.go so they leave the production binary (all
callers are same-package tests): the eight ws test-client constructors
and voice/E2EE setters from ws/client.go, admin.SetBackupBaseDir
(new admin/export_test.go), api.SecurityHeaders (test-only wrapper;
production uses SecurityHeadersWithTLS — docs/api.md updated to the
real name). Client.getVoiceJoinToken/setVoiceChID inlined into their
existing ForTest wrappers; TestSetVoiceChID_* self-tests deleted.
Kept after verification: updater.SetBaseURL (11 cross-package test call
sites) and telemetry.resetAppMetricsForInit (live under -tags otel —
untagged deadcode false positive).
Full gate green: gofmt/vet, 4 build-tag variants, full suite, deadlock,
race.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Closes audit finding A-2026-07-16, two defects in the same rule:
- permissions.HasServerPerm (admin bypass OR all-of bit test) replaces
the hand-rolled copies in api.RequirePermission (whose raw test was
any-of for multi-bit masks) and ModerationService.requireBanPermission.
RequirePermission's doc comment now states the scope contract: role
bitfield only, channel overrides deliberately not consulted.
- PermissionService.getOrPopulate and ChannelService.ListVisibleChannels
no longer substitute an empty override map when
GetAllChannelPermissionsForRole errors. That silently dropped every
channel-level deny — and the permission cache then served the degraded
snapshot for permCacheTTL (30s) across ~25 callers. Both fail closed
now; admins skip the fetch entirely (they bypass channel checks).
- PermissionService.HasChannelPerm delegates to Checker.HasChannelPermBatch
and MessageService.GetAccessibleChannelIDs to VisibleChannelIDs — the
missed fifth D9 site, making that closure true rather than aspirational.
- AuthMiddleware rejects a dangling role_id (GetRoleByID returns nil,
nil) with 401 instead of putting a nil role in the request context.
Locked by failing-first tests: override-fetch-error denies (cached and
uncached paths), admin-outage skip, multi-bit all-of, channel allow
override must not grant a server-wide route, 403 locks on both
RequirePermission routes, dangling-role 401.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The per-username brute-force lockout keyed on the raw request username while GetUserByUsername matches COLLATE NOCASE, so case variants (admin/Admin/ADMIN) each got an independent 9-attempt bucket, multiplying allowed guesses per account. Lowercase the username before building the login_user_fail/login_user_lock keys so all casings share one bucket. (Security scan F1)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
nhooyr.io/websocket now resolves to github.com/nhooyr/websocket-old and its
README is a one-line deprecation pointing at coder/websocket. Its last three
releases (v1.8.15-17) all shipped on 2024-08-10 as the redirect; the fork has
shipped through 2026-06-15.
The version number decreases (v1.8.17 -> v1.8.15) because both paths tagged in
the same space, but the coder release is ~2 years newer. Import path only; the
9 API symbols used are unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The client held the Klipy key in VITE_KLIPY_API_KEY, which Vite inlines into
the shipped bundle by design — a build variable can never hold a secret. Move
the integration behind the server:
- New authenticated GET /api/v1/gif/search and /api/v1/gif/trending. The key
comes from the new `gif.api_key` config section (koanf,
OWNCORD_GIF_API_KEY) and never leaves the server.
- Default-off: with no key, both endpoints return 503 GIF_DISABLED so clients
can hide the picker instead of showing a broken one. Auth is checked first,
so anonymous callers cannot probe whether a key is configured.
- Outbound call reuses the existing SSRF-guarded dialer (exported as
plugin.GuardedDialContext) rather than a bare http.Get: resolve once,
reject private/loopback/link-local/CGN, dial only vetted IPs. Redirects are
not followed and the response body is size-capped.
- Only id/title/media_formats.{tinygif,gif}.url are forwarded — decoding into
the narrow struct is the allowlist, so an upstream that echoed the key
could not leak it. Upstream errors become a generic 502 and the key is
redacted from anything that reaches the logs.
- Dedicated `gif:` rate-limit bucket (30/min per IP) so debounced search
traffic cannot exhaust the shared bucket used by password/TOTP endpoints.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Audit writes stay best-effort — a LogAudit failure must never fail or abort
the request — but a failed write must no longer be silently discarded. Add
db.WriteAudit(auditor, actor, action, targetType, targetID, detail), which
logs a failed write with actor/action/target context (never the detail
string, which may be sensitive) and never propagates the error.
The Auditor interface is satisfied structurally by both *db.DB and the
service-layer Store, so api/admin/ws/service all reach the helper without an
import cycle. Converts all ~26 call sites from `_ = LogAudit(...)` (and the
two backup handlers' inline `if err` blocks) to db.WriteAudit. Pinned by
db/audit_test.go: failure logged and not propagated, success logs nothing,
detail never leaks.
Resolves the repo-wide LogAudit policy question flagged by the D8 note.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Deletes Server/store (SQLiteStore, MemStore, the composed Store
interface) and collapses to a single sqlc-backed db package, executing
the prior audit's P4 "single data layer" direction (finding #6).
SQLiteStore was a pure pass-through to *db.DB, so consumers now depend
on narrow interfaces that *db.DB satisfies directly:
- service.Store (service/datastore.go, renamed from store/store.go)
- ws.EventStore (ws/eventstore.go)
- plugin.PluginStore (plugin/pluginstore.go)
The event- and plugin-KV methods that lived in the store's SQLite
implementation move into the db package (db/event_queries.go,
db/plugin_queries.go), keeping their raw-SQL form.
Tests: the MemStore-based unit tests now run against a real in-memory
SQLite db opened per-test with migrations applied, via package-local
seed helpers. Fault-injection tests embed a real *db.DB and override the
single method under test, preserving error-path coverage. Full server
suite and sqlc-verify are green.
Docs: audit finding #6 and A-2026-07-06 marked resolved; decisions D3
updated; architecture server.md / data-model.md diagrams and prose
updated to the api -> service -> db layering.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UA17KPvqGBX3XbXYnMf1rA
Protocol codegen (decision D4, audit A-2026-07-08):
- Add docs/protocol-schema.json as the real single source of truth for
WS message-type constants, making the long-standing 'generated from'
comment in both constant files true.
- Add Server/scripts/genprotocol, a generator emitting both
Server/ws/message_types.go and Client .../lib/protocolTypes.ts
(constants byte-for-byte value-identical to before; only headers,
ordering alignment, and provenance comments changed).
- Add make protocol-generate / protocol-verify and wire protocol-verify
into CI next to sqlc-verify.
Quick wins (decision D8):
- admin: log LogAudit write failures in the backup handlers instead of
discarding them (prior audit #10).
- api: fix self-contradictory upload Cache-Control to 'private,
no-cache' per remediation plan W3-4; drop the now-unused
fileCacheMaxAgeSeconds constant; update test.
- ws: route the hub settings cache through db.GetSetting instead of
inline SQL.
- ws: fix a latent data race — main.go wires SetEventPersister and
SetEventStore after NewRouter has already started the hub Run loop,
which reads those fields on the broadcast/replay paths. They (and
pluginSink, which one test sets post-Run) are now atomic pointers;
the remaining pre-Run-only setters reject late calls with an error
log instead of racing silently.
Update the audit closure table and decisions doc statuses accordingly.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UA17KPvqGBX3XbXYnMf1rA
With trusted_proxies covering client networks (e.g. 10.0.0.0/8 over LAN
clients), the right-to-left XFF walk skipped every entry, exhausted, and
fell back to the proxy's RemoteAddr — collapsing all clients into one
rate-limit/lockout bucket, so one user's failed logins locked out
everyone. On exhaustion the walk now returns the leftmost valid entry
(furthest-upstream hop), the best distinct per-client key such a config
allows. An untrusted RemoteAddr still never gets its headers honoured.
Also (W3-3): the CIDR list parses once per request instead of once per
XFF candidate, config load warns about invalid CIDR entries at startup
(a silently skipped entry silently un-trusts the proxy), and the sample
config documents that trusted_proxies must list only proxy hops.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
UpdateUserPassword commits first; when DeleteOtherSessions then errored the
handler returned 500 and skipped the audit row — telling the user the
change failed while the new password was already live, walking them into
retrying with a dead password and tripping the confirm lockout. The
committed change now always audits and reports success; revocation gets
one bounded compensating retry, and a persistent failure surfaces as a
200 + warning (sessions_revoked count) the client can show.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The empty-prefix middleware shared per-IP buckets with verify-totp,
password change, and the sensitive endpoints, so a client's 30/min
auto-poll could 429 its own user's 2FA or password change. Dedicated
"client_update:" prefix, mirroring "livekit_proxy:".
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
requireBanAuthority (BAN_MEMBERS + role hierarchy) was wired only into
ModerationService.BanUser/UnbanUser — which had zero production callers.
The live path, handlePatchUser, ran a raw UPDATE with no hierarchy check,
so any admin-panel actor could ban an equal- or higher-ranked user,
including the owner. The ban/unban branch now calls the service (dead code
becomes THE code — ban path 1 of 3 consolidated), which also audits as
user_ban/user_unban, keeping the historical audit vocabulary.
Authorization now runs in permission → existence → hierarchy order: an
actor without ban authority sees Forbidden, never NotFound, so the ban
path cannot enumerate user ids. The role+ban transaction is gone — the
ban leg lives in the service, runs first, and a refusal returns before
the role change executes, so a rejected ban never half-applies a PATCH.
MemStore gains honest BanUser/UnbanUser so the matrix is testable.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Publish every release to the public OwnCord-releases repo with a full
source snapshot (AGPL section 6 for binary recipients), and make the
updater's repo coordinates configurable (github.owner/github.repo),
defaulting to OwnCord-releases. Both the server self-update and the
/client-update chain follow the new default, so deployed servers keep
updating after the source repo goes private.
The publish step fails closed: a private source repo with no
RELEASES_REPO_TOKEN aborts the release instead of silently shipping
binaries with no public source or update feed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Applies fixes for 20 adversarially-verified findings from a whole-codebase
security review (server side). All Go build-tag variants build, `go vet` is
clean, and the suite passes (the sole failing test, ws TestEmitEvents, is a
pre-existing nil-harness failure unrelated to these changes).
High severity:
- auth: close TOCTOU in TOTP verify rate-limit by recording each attempt
atomically up-front (was Check-then-Allow), restoring the per-user
brute-force cap.
- plugin: enforce the CPU/time budget on every WASM guest call via a
WithTimeout context (WithCloseOnContextDone interrupts runaways); the
configured budget was previously parsed but never applied.
- api/waf: inspect request bodies for chunked (ContentLength==-1) requests
so the SQLi/XSS/RCE body rules can no longer be bypassed.
- ws: rate-limit voice_join/voice_leave and voice_e2ee announce/offer, which
fan out to every participant and could force mass disconnects.
Medium severity:
- api: run bcrypt on the unknown-user login path (no || short-circuit) to
remove the timing-based username-enumeration oracle.
- ws: verify LiveKit webhooks via the SDK receiver so the signature is bound
to the body hash (kills forgery/replay).
- authz: require READ_MESSAGES for reactions and for plugin-command
broadcasts; route the latter through RequireChannelAccess.
- api: cache the client-update signature fetch and rate-limit the endpoint.
- service: propagate DeleteOtherSessions failure from ChangePassword instead
of silently reporting success.
- api: trust the rightmost non-proxy X-Forwarded-For entry, not the
client-controllable leftmost one.
- plugin: route auto-registered commands through the conflict-checked
RegisterCommand; pin the DNS-validated IP for host_http dials
(DNS-rebinding TOCTOU).
- api: mark access-controlled downloads private/no-cache + Vary: Origin.
Low severity:
- auth: fail closed when a fully-shaped TOTP ciphertext fails GCM auth
(was returning the ciphertext as plaintext).
- api: apply the livekit-proxy path allowlist to WebSocket upgrades too.
- service: verify attachment ownership before linking (IDOR).
- admin: bound the bootstrap setup invite (5 uses / 24h); re-verify the
update binary hash immediately before rename+spawn (TOCTOU).
- service: require BanMembers + role hierarchy for moderation ban/unban.
chore: stop tracking the stray Server/owncord-server.exe build artifact.
Test infra: add uploader_id to the hand-rolled ws test attachment schemas and
make MemStore.GetAttachmentByID a no-op lookup, matching production/DB behavior.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
contextcheck: add ctx context.Context as first param to ListVisibleChannels,
BlockUser, CreateDM, CreateInvite, UpdateProfile, SendMessage; pass r.Context()
from HTTP handlers and ctx from WS handler; replace context.Background() in
telemetry spans with the propagated ctx.
errcheck: suppress justified Close() errors — defer func(){ _ = rows.Close() }()
in sqlite_events.go (idiomatic; rows.Err() checked), _ = resp.Body.Close() in
host_http.go (body fully consumed), _ = f.Close() in host_ui.go (read-only fd).
gocritic/rangeValCopy: rewrite for _, ch := range all (line 60) to indexed loop
in ChannelService.ListVisibleChannels to avoid 144-byte per-iteration copy.
Resolves the parallel Phase B/C work that landed on the sister branch
while this branch was in review. Both branches independently implemented
real OTel + Wazero runtimes; this merge keeps the best of each.
Conflict resolution
- Server/plugin/sandbox_wazero.go: rewritten as a hybrid. Keeps the
HEAD lifecycle (eager `platformInit` with WASI preview-1, explicit
`platformDeactivate` per-instance, runtime closed via Registry.Close)
AND adopts the sister branch's richer artefacts:
* `WithMemoryLimitPages` actually enforces `cfg.MaxMemoryMB`,
* the JSON-over-linear-memory ABI
(`allocate` / `command_dispatch(ptr,len) → (ptr,len)` /
`deallocate`),
* `listExportedCommands` auto-binds commands the plugin exports
via `list_commands` at activation time (capability-gated).
- Server/plugin/registry.go: kept the HEAD `activate()` snapshot
pattern (read `runtimePlatform` under RLock, pass into
`activateWithRuntime` as a parameter) so a concurrent Close can't
race the wazero call. Sister branch's LoadAll stale-staging cleanup
and UninstallPlugin on-disk dir removal came in via auto-merge.
- Server/telemetry/telemetry_otel.go: kept the HEAD implementation
(race-fixed AppMetrics rebind, uint64 overflow guard, idempotent
shutdown, trace-provider cleanup on prom failure) and wired in the
sister branch's `OTLPInsecure` config field for plaintext gRPC opt-in.
- Server/go.mod: accepted sister branch's `BurntSushi/toml v1.6.0`
for the new TOML manifest support.
- Client/tauri-client/vitest.config.ts: union of both globs
(`tests/**/*.test.ts`, `src/**/*.test.ts`, `src/**/*.test.tsx`).
- PHASE_BC_LOCAL_TODO.md: combined the two checkbox histories;
TOML manifest, hello.wasm fixture, and OTLPInsecure are all marked
done now.
Sister branch additions accepted via auto-merge
- Server/plugin/examples/hello/{hello.wasm,main.go}: precompiled 925
KiB TinyGo plugin with the full ABI (allocate, deallocate,
list_commands, command_dispatch, on_event).
- Server/plugin/manifest_{toml,nottoml}.go: TOML manifest parser
behind the wazero build tag, JSON fallback elsewhere.
- Server/plugin/loader.go: prefers `plugin.toml`, falls back to
`plugin.json`.
- Server/api/plugins_handler.go: structured error responses + slog.
- Server/main.go, Server/config/config.go: OTLPInsecure plumbing,
defaults polish.
- docs/{contributing.md,server-configuration.md}: documentation
updates.
Test status
- `go build` passes on default, -tags otel, -tags wazero, and
-tags otel,wazero.
- `go vet` passes on every tag combination.
- `go test ./...` passes on default and on -tags otel,wazero.
- Client: `npx tsc --noEmit` clean; vitest 3188/3188 across 112 files.
https://claude.ai/code/session_01AZni6CDSQeu67WSWY1YCDX
- sandbox_wazero.go: explicitly discard plugin stdin via WithStdin to
prevent WASM modules from reading the server process's stdin fd
- channel_handler.go: replace ErrInternal message pass-through with
generic 'an internal error occurred' — full error stays server-side
in slog.Error only
- invite_handler_test.go: update assertions to expect generic message
- service/user.go: fix ChangePassword docstring (no old-password verification)
- service/user.go: RevokeSession now maps db.ErrNotFound→ErrNotFound and
all other store errors→ErrInternal, preventing internal failures from
masquerading as 404s
- plugin/loader.go: update scanPluginDirectory comment to reflect fail-fast
behavior; fix Lstat comment wording
- db/queries/sqlite/events.sql: CAST COALESCE result to INTEGER so sqlc
generates int64 instead of interface{}
- api/plugins_handler.go: log install error server-side and return sanitized
structured JSON response instead of raw err.Error()
- .github/workflows/ci.yml: remove continue-on-error from tag build steps
so tag boundary drift fails CI"
Agent-Logs-Url: https://github.com/J3vb/OwnCord/sessions/6635420c-af26-4cc0-9397-5e5b37887437
Co-authored-by: J3vb <192430104+J3vb@users.noreply.github.com>
Phase B Step 8 (OpenTelemetry) and Phase C Step 9 (Wazero plugin runtime)
were structurally scaffolded but the tagged builds were placeholders that
errored at runtime. This commit lands the real implementations behind the
existing build tags, plus three review passes worth of fixes across the
plugin admin handler, plugin registry, telemetry adapter, and Solid client.
Telemetry (Phase B Step 8)
- Add real go.opentelemetry.io/otel{,/sdk,/exporters/{prometheus,otlp...}}
modules to go.mod plus contrib/instrumentation/net/http/otelhttp.
- Replace the telemetry_otel.go skeleton with a working Provider that
wires Prometheus + OTLP/gRPC exporters, otelhttp middleware, span and
meter adapters, and an idempotent Shutdown.
- AppMetrics cache is now reset *before* SetGlobal to close a race where
a concurrent NewAppMetrics() could observe a swapped provider but read
stale no-op instruments.
- Init releases the trace provider on a later prometheus exporter
failure so Init never leaks gRPC connections.
- convertAttrs handles int32/uint/uint32/uint64/float32 explicitly;
uint64 values that exceed math.MaxInt64 fall back to a STRING attr
rather than wrapping into a negative int64 and corrupting metrics.
- Tests under -tags otel cover the prometheus scrape, span lifecycle,
histogram recording, shutdown idempotency, AppMetrics rebind, and
the uint64 overflow fallback.
Plugin runtime (Phase C Step 9)
- Add github.com/tetratelabs/wazero v1.11.0 to go.mod.
- platformInit creates a shared wazero.Runtime with WASI preview1
pre-instantiated; activateWithRuntime compiles + instantiates each
plugin module under that runtime; platformDeactivate closes per-
plugin modules without tearing down the runtime.
- DisablePlugin now calls platformDeactivate so the wazero module is
freed immediately instead of leaking until registry Close.
- activate() captures runtimePlatform under r.mu.RLock and passes it as
a parameter to activateWithRuntime; the call no longer re-reads the
field, closing a race with concurrent Close.
- invokeCommand calls the plugin's command_dispatch export when
present; missing/broken exports return a user-facing diagnostic
instead of crashing the dispatcher.
- Tests under -tags wazero cover registry creation, module compilation,
re-enable after disable (verifies the leak fix), close-twice safety,
invalid wasm rejection, and DispatchCommand with a missing export.
Fixture is a 41-byte embedded add.wasm; no external asset required.
Plugin admin handler hardening
- /api/v1/admin/plugins/install now rejects uploads whose multipart
Content-Type is not application/zip|x-zip-compressed|octet-stream
(415) and uploads whose body lacks the PK\\x03\\x04 / PK\\x05\\x06
zip magic (400). The 16 MiB cap and registry-side zip-slip / symlink
/ size-bomb defences are still applied as before.
- New plugins_handler_test.go covers list-empty, install-503-when-nil,
content-type rejection, magic rejection, happy path, lifecycle 503,
invalid id, and isZipContentType / hasZipMagic helpers.
Solid client (Phase B Step 6) cleanup
- vitest.config.ts now wires vite-plugin-solid and broadens the test
glob to include src/**/*.test.tsx so Badge.test.tsx is actually
discovered (it was silently skipped).
- pluginBridge.ts targets postMessage at window.location.origin
instead of "*", and exposes a destroy() that detaches the message
listener and clears mounted frames.
- solidMount.ts imports the JSX type from "solid-js" instead of
"solid-js/web" (the latter does not re-export it), unblocking
npx tsc --noEmit.
Build/test status
- go build succeeds on default, -tags otel, -tags wazero, and
-tags otel,wazero.
- go test passes on every tag combination across telemetry, plugin,
api, ws, service, store, and the rest of the tree.
- Client: npx tsc --noEmit clean; vitest 3188/3188 across 112 files.
PHASE_BC_LOCAL_TODO.md is updated to mark the OTel modules + real Init,
the wazero module + real platformInit, and the test coverage that
landed in this commit as completed.
https://claude.ai/code/session_01AZni6CDSQeu67WSWY1YCDX
Clean sweep of every actionable item from the two Copilot review passes
on head 59ae4d8. Grouped by severity:
─── Crash / security (must-fix) ─────────────────────────────────────
1. main.go:140 — telemetryShutdown nil panic.
telemetry.Init can return (nil, err) on the -tags otel skeleton
path; the deferred closure would then call a nil function. Normalise
to a no-op shutdown when Init errors so the defer is always safe.
2. api/upload_handler.go — permSvc nil deref.
MountUploadRoutes + handleServeFile dereference permSvc on every
authenticated file request. Add a fail-fast panic at mount time so
the misconfiguration surfaces at wiring, not on the first 500.
Update upload_handler_test.go to pass a real PermissionService built
on the test DB (the existing tests were missing the argument entirely,
which meant the package wouldn't compile — this fixes the real bug
Copilot flagged).
3. ws/event_persister.go — NewEventPersister nil EventStore panic.
run() dereferences p.store on every flush. Panic at constructor
time instead so the crash happens once at startup rather than
minutes later in a background goroutine.
4. plugin/host_ui.go — serve-time symlink check.
rejectSymlinksUnder only runs at install time, so a symlink created
post-install (accidental or malicious) would be followed by
http.ServeFile and leak host files. Add an os.Lstat + ModeSymlink
check + IsRegular check to AssetHandler on every request. Cheap
relative to the file read and closes the TOCTOU window.
─── Correctness / observability (should-fix) ───────────────────────
5. ws/deps.go:77 — requirePerm hides misconfig as FORBIDDEN.
Previously, nil database, nil perms, or a GetRoleForUser error all
returned ErrCodeForbidden with the same message, making operator
failures indistinguishable from legitimate permission denials.
Split the branches: misconfig + DB error now return ErrCodeInternal
with a server-side slog.Error so operators see the real problem;
FORBIDDEN is reserved for the actual permission-bit check.
6. telemetry/metrics.go — ServiceCallDurationMs renamed to Sec.
Field name said "Ms" but the instrument name was
`service_call_duration_seconds` with unit "s". Renamed the field
and updated all 8 service-layer callers so the struct field and
metric semantics match.
7. ws/event_persister.go — flushEvy typo → flushEvery.
Renamed the field and the one call site in run().
─── Comments out of sync with code ──────────────────────────────────
8. plugin/loader.go — Stat vs Lstat comment.
The comment claimed "Stat (not Lstat)" but the code correctly uses
os.Lstat to detect symlinks. Updated the comment to match the code;
the code was already right.
9. telemetry/telemetry_otel.go — compile claim wrong.
Comment said the file would fail to compile without the upstream
OTel modules, but the skeleton deliberately avoids importing them
and Init returns a runtime error instead. Updated the comment to
reflect actual CI behaviour (the -tags otel build step passes
today but doesn't exercise real telemetry).
─── Nit / polish ────────────────────────────────────────────────────
10. ws/event_pruner.go — startup delay magic constant.
Hard-coded time.Minute made the "run shortly after startup"
behaviour untestable (a test with a 100ms interval would still
wait a full minute). Cap the startup delay by the interval:
min(interval, time.Minute). Documented via a new `maxStartupDelay`
constant.
11. ws/event_pruner_test.go — new file.
Unit coverage for runPrune cutoff correctness, error swallowing,
StartEventPruner nil-store short-circuit, ctx cancellation, and
the interval-bounded startup delay from fix#10. Uses a fakeEventStore
stub that records every prune call and signals the first one so
tests don't sleep.
─── Verification ────────────────────────────────────────────────────
gofmt -l clean. No network access in sandbox so `go vet` and `go test`
could not run; the changes are local and surgical and every touched
file compiles in isolation against the existing signatures.
https://claude.ai/code/session_01UsBsQW2YiA2usk9pnJjAWk
Final in-sandbox completeness pass. Five focused pieces; the remaining
items in PHASE_BC_LOCAL_TODO.md after this commit are all genuinely
local-only (toolchain, network, native deps).
Test coverage (the biggest gap from prior reviews)
- Server/plugin/manifest_test.go — pluginNameRegexp accept/reject table,
validateRelativePath table, oversized version, unknown permission.
- Server/plugin/host_http_test.go — hostAllowed dot-boundary suffix,
empty-entry rejection, case insensitivity, FQDN trailing dot. ipAllowed
table over loopback, RFC1918, RFC4193 (ULA), RFC6598 (CGN), link-local,
multicast, unspecified — both v4 and v6 — plus public-IP accept cases.
- Server/plugin/loader_test.go — rejectSymlinksUnder catches direct and
nested symlinks; scanPluginDirectory rejects a plugin whose entrypoint
is a symlink. Skipped on Windows where symlink creation needs elevation.
- Server/plugin/host_ui_test.go — AssetHandler serves declared files,
rejects undeclared files (404), rejects path traversal, supports nested
asset paths.
- Server/ws/hub_seedseq_test.go — SeedSeq monotonic, never-backwards,
concurrent CAS safety, integration with nextSeq.
- Server/ws/extract_event_type_test.go — table covering happy paths,
control char rejection, escaped quote rejection, length cap (64),
empty/missing/non-JSON inputs.
Plugin install endpoint (closes a real feature gap)
- Server/plugin/registry.go — InstallFromZip extracts a plugin .zip into
a staging directory under cfg.Directory, validates it zip-slip safe
(cleaned-path Rel check), refuses non-regular entries, refuses
symlinks, caps compressed at 16 MiB and uncompressed total at 64 MiB
(each file gated by io.CopyN against the remaining budget). Manifest
is parsed at the staged root, then atomically renamed into the
canonical plugin directory and registered via the existing
installFromDisk path.
- Server/api/plugins_handler.go — POST /install accepts multipart with
one "plugin" file part, http.MaxBytesReader caps the request body,
io.LimitReader caps the in-memory buffer, calls Registry.InstallFromZip,
returns 201 with the new plugin name. The endpoint inherits the Pass 2
admin auth + IP gate (mounted under r.Use(admin.RequireAdminAuth)).
Protocol surface
- Server/ws/serve.go — buildAuthOK now takes replaySource and includes
it in the auth_ok payload as "replay_source": "none" | "buffer" | "db".
Two call sites updated: reconnect path passes the existing local,
fresh-connect path passes "none". Test export updated to pass "none".
CI build-tag matrix
- .github/workflows/ci.yml — three new steps inside server-build-test
build the server with -tags otel, -tags wazero, and -tags otel,wazero.
All three are continue-on-error: true until the upstream OTel and
wazero modules land in go.mod (tracked in PHASE_BC_LOCAL_TODO.md).
Once they do, dropping continue-on-error converts the steps into
hard CI gates against tag-boundary drift.
Documentation
- CHANGELOG.md — new root-level file with curated entries for Phase B,
Phase C, security, and behavioural changes operators must know about
(notably event_persistence.enabled = true by default).
- PHASE_BC_LOCAL_TODO.md — ticks off the install endpoint, the
replay_source field, and the existing event_persistence defaultYAML
entry. The remaining items are toolchain-bound.
After this pass, the in-sandbox completeness ceiling is reached.
Everything still pending requires Go 1.25 toolchain, npm install,
real OTel SDK + wazero modules, sqlc, postgres backend impl, or
tinygo.
https://claude.ai/code/session_01UsBsQW2YiA2usk9pnJjAWk
Phase B + C review pass: critical security and correctness fixes.
Security
- S1: plugin admin endpoints now require admin.RequireAdminAuth in addition
to AdminIPRestrict. Previously a LAN attacker on the allowed CIDR could
list/enable/disable/uninstall plugins without a session.
- S2: rewrite plugin HTTPDo allowlist with proper net/url parsing. Empty
entries are ignored, suffix matches require a dot boundary, and a custom
Dialer rejects loopback / RFC1918 / link-local addresses to close the
DNS-rebinding TOCTOU window. Redirects re-validated.
- S3 + #9: manifest Name pinned to ^[a-z0-9][a-z0-9_-]{0,63}$, Entrypoint
and UI tab assets validated against absolute / "..", NUL byte, backslash
and non-canonical paths. Asset handler hardened with filepath.Rel check
for symlink and prefix-without-separator escapes.
- S5: pluginBridge postMessage handler ignores the pluginId in the message
body and uses an e.source -> contentWindow lookup instead, defeating
spoofed messages from same-origin scripts.
- S8: HTTPDo body capped at 5 MiB via io.LimitReader, redirects bounded
to 5 hops.
Correctness
- Critical seq alignment: PersistEvent now takes the hub-assigned seq as a
required parameter so the events table row seq always matches the wrapped
payload seq. Hub seeds its in-memory atomic counter from MAX(events.seq)
on startup. Drops in the persister queue no longer mis-align row vs
payload seq.
- #1: live plugin.Registry constructed in main.go BEFORE NewRouter and
threaded through; admin handler is no longer wired with nil.
- #3: EventPersister.Stop is now safe to call without a prior Start by
tracking a started flag — previously deadlocked waiting on done.
Wiring
- NewRouter signature gains *plugin.Registry; two test callers updated.
- admin.RequireAdminAuth exported as a thin wrapper over the existing
package-private adminAuthMiddleware.
- sqlc query templates updated for the new PersistEvent + GetMaxEventSeq
contracts (sqlite + postgres).
https://claude.ai/code/session_01UsBsQW2YiA2usk9pnJjAWk
- upload_handler.go: uses PermissionService.HasChannelPerm instead of
the deleted hasChannelPermREST helper
- profile_handler.go: delegates to UserService for profile updates,
password changes, session listing, and session revocation
- Remove hasChannelPermREST from channel_handler.go (no longer needed)
- UserService.UpdateProfile now returns ErrConflict on duplicate username
https://claude.ai/code/session_01CBFF3r84ywkJRWwuqw8zD8
Critical fixes:
- Move svc creation in router.go above MountInviteRoutes/MountChannelRoutes
(was used before definition — compile error)
- Restore hasChannelPermREST in channel_handler.go for upload_handler.go
(was removed but still referenced — compile error)
Permission cache invalidation:
- Add PermissionInvalidator interface to admin package
- Wire through NewHandler → NewAdminAPI → handlePatchUser
- Call InvalidateUser(userID) after role changes in admin panel
- Update all admin test files to pass nil as new parameter
Also clarifies WithTx documentation for SQLite single-writer semantics.
https://claude.ai/code/session_01CBFF3r84ywkJRWwuqw8zD8
- Add UserService, DMService, InviteService, BlockService
- Migrate REST handlers (channel, DM, invite, block) to use services
- Remove block_handler.go (merged into dm_handler.go)
- Update all services to accept store.Store instead of *db.DB
- Router creates SQLiteStore and passes to service.New()
Handlers are now thin HTTP adapters: parse request → call service →
map error → write JSON. All business logic lives in the service layer.
https://claude.ai/code/session_01CBFF3r84ywkJRWwuqw8zD8
Introduce Server/service/ package with MessageService, ChannelService,
and PermissionService that encapsulate business logic previously
scattered across REST and WS handlers. The PermissionService adds
per-user in-memory caching with TTL-based expiry to eliminate
per-message DB round-trips at scale.
Services are wired into the WS hub via deps structs (strangler-fig
pattern) — existing handlers continue to work unchanged, with service
references available for incremental migration.
https://claude.ai/code/session_01CBFF3r84ywkJRWwuqw8zD8
- Remove unused `ver` param from handleHealth and handleInfo (version
was intentionally removed from unauthenticated endpoints per C-2)
- Rename decodeBase64Loose to validateBase64Loose returning only error,
matching actual usage (all callers only validate, never use the bytes)
- I-1: Add key holder election in Hub (lowest userID per channel); reject
non-key-holder voice_e2ee_offer with NOT_KEY_HOLDER error
- I-2: Accept raw (unpadded) base64 in E2EE announce/offer handlers via
decodeBase64Loose fallback
- I-6: Copy E2EE public key value while h.mu.RLock is held in getClientE2EEPubKey
- I-7: Lower loginRateLimitPerMinute from 60 to 5
- C-1: TOCTOU fix — target channel check held under same lock as client lookup
- C-2: Include is_key_holder bool in voice_token payload so client knows
whether to initiate key distribution
- M-5/M-6: Add ErrCodeBadPayload/ErrCodeNotKeyHolder error constants
- Fix pre-existing api build errors: block_handler.go getUserFromContext,
router.go RequirePermission arg count
- Add user_blocks table to all test DB schemas (ws, api DM)
- Add voice_e2ee_test.go and constants_test.go covering all fixes
Update LiveKitSession tests to use _state discriminated union instead of
old flat field names (room, currentChannelId, latestToken, etc.) removed
in the state machine refactor. Also fix renderers.test.ts URL resolution
by setting a server host in beforeEach so isSafeUrl can parse relative
attachment URLs in jsdom. Stage all four Go test files so the CI Go job
runs them.
Additionally fix a regression in connectAndSetup's finally block: when a
pendingJoin is queued during a stale-join abort, preserve the connecting
state so handleVoiceToken's drain loop can pick it up rather than losing
it by resetting to idle.
