Record the isolated Jellyfin 12 live evidence and the server-plugin-only no-go. Remove the discovery snapshot invalidated by Jellyfin's actual playback call path, scope the guide cache to connection settings, and fail closed for dangling mirror sources.
Refs #357
Co-Authored-By: OpenAI Codex <codex@openai.com>
Pin discovery/open to a short schedule-identity snapshot without sharing mutable Jellyfin media sources, gate the authorization-incomplete lab probe behind an explicit opt-in, and add focused plugin tests to the main solution. Redact remote URLs and serialize OpenAPI ApiKey requirements correctly.
Refs #357
Co-Authored-By: OpenAI Codex <codex@openai.com>
Add an authenticated playback-source endpoint for player-owned playback and document the Jellyfin Live TV proof-of-concept boundary. Cache guide snapshots for the probe plugin while leaving native Jellyfin source handling in control.
Refs #357
Co-Authored-By: OpenAI Codex <codex@openai.com>
Add a Jellyfin 12 rc2 ILiveTvService probe that reads ErsatzTV guide/source data and returns native Jellyfin media sources while preserving the LiveTvChannel session item.
Refs #357
Co-Authored-By: OpenAI Codex <codex@openai.com>
Retrigger CI after the existing Libraries scan-progress test failed once remotely but passed six consecutive local runs at the same documentation-only PR head.
Refs #355
Co-Authored-By: OpenAI Codex <codex@openai.com>
Retrigger CI after the migration job failed in Setup .NET while cleaning a missing runner-cache file, before any migration command executed.
Refs #355
Co-Authored-By: OpenAI Codex <codex@openai.com>
Record the tested GPT-5.4-Mini low-effort worker command, current spawn-agent limitation, and unsupported legacy model trap so future sessions can route bounded work without rediscovery.
Refs #355
Co-Authored-By: OpenAI Codex <codex@openai.com>
Clarify that enabled web_search makes a cheap worker tool-bearing even when the selector intends to use only Gitea, reserve minimal for tool-free supplied-evidence synthesis, and require hard evidence/output caps.
Co-Authored-By: OpenAI Codex <codex@openai.com>
Exercise default-deco persistence and errors plus the complete action, add-dialog, and edit-field kind matrices identified by cold review.
Refs #245
Co-Authored-By: OpenAI Codex <codex@openai.com>
Disable and gate remote connection draft inputs while a save owns the current revision. Exercise the real App-owned route transition after a successful save so the regression proves the editor unmounts without a dirty prompt.
Refs #344
Co-Authored-By: OpenAI Codex <codex@openai.com>
Cold review of PR #331:
- Move the ZAP-container reap from an inline post-scan line into cleanup() (the
EXIT trap) so a SIGINT/timeout kill mid-scan can't leave it running.
- semgrep absent now prints a stderr WARN instead of silently skipping the SAST
pass (no false confidence for a release gate).
refs #314
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
scripts/security-scan.sh: boots a THROWAWAY container from the image under test
(fresh config volume; never the deployed prod/test container — the active scan
attacks write endpoints), reads its machine key, and runs an authenticated OWASP
ZAP api-scan that imports /openapi/v1.json (all 160 /api/v1 ops) and injects
X-Api-Key on every request via a replacer rule so it reaches the
[RequiresAuthentication] + RequireKeyForReads surface — then a semgrep SAST
cross-check. Wrapped in `timeout` because zap-api-scan can hang in post-scan
cleanup after the report is already written.
docs/ci-cd.md: new 'Security scanning' section (out-of-ecosystem black-box gate,
run on the docker host per-release like migration-smoke, not a per-PR CI job) +
the Microsoft.OpenApi 2.7.5 pin note in dependency management.
refs #314#197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Microsoft.AspNetCore.OpenApi 10.0.2 + Scalar.AspNetCore pull Microsoft.OpenApi
2.0.0 transitively — High severity (stack overflow parsing a circular $ref;
fixed in 2.7.5). Direct-pin 2.7.5 in CPM + a direct ErsatzTV PackageReference so
the override actually resolves (the #8 SQLitePCLRaw transitive-override pattern).
Build clean (transformers compile unchanged against 2.7.5), v1.json byte-identical
(no OpenAPI-output change), ErsatzTV + ErsatzTV.Tests both vuln-clear, 198
OpenAPI/contract/security tests green.
refs #314#8
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold review (PR #325) caught that removing the unreachable 404 from
BlockController/TemplateController.CreateGroup left ApiErrorResponseMetadataTests
still asserting those ops document 404 -> red Build & test. Removed those two stale
assertions. For spec consistency, extended the trim to the two OTHER create-group
actions carrying the same unreachable 404 (DecoController, DecoTemplateController --
their Create*GroupHandler only do a duplicate-name AnyAsync -> 422, never a lookup
that 404s). Net: all four CreateGroup 404 assertions removed (422 siblings kept),
both controllers trimmed, v1.json regenerated (4 unreachable 404 blocks gone total).
PlaylistController.CreateGroup already carried no 404.
Refs #172
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Clears the still-live findings from #172 (verified against main; #2/#4/#7 and the
auth/search/Trakt tail were already deliberate-documented or fixed since 2026-07-07).
- Null/empty Name → 500 (10 create/replace handlers). Block/Template/DecoTemplate/Deco
Create+Replace/Update + UpdateFFmpegProfile did `request.Name.Length > 50` on a
client-nullable string → unhandled NullReferenceException → HTTP 500 (no global
exception filter). Now `string.IsNullOrWhiteSpace(request.Name) || .Length > 50` →
422; also rejects empty/whitespace names, matching the group-create handlers'
NotEmpty behavior. CreatePlaylist coalesces null→"" at the DTO so it was an
empty-name persist, not a 500; guarded the same way.
- ReplaceTemplateItems overlap validation iterated with an `item == otherItem`
record value-equality skip, so two exact-duplicate items were value-equal and
bypassed the intersection check (both persisted). Now index-based (i != j) so
duplicates register as a self-intersection and are rejected 422.
- Trimmed the unreachable 404 ProducesResponseType from POST /api/blocks/groups and
POST /api/templates/groups (a create has no parent lookup that can 404); v1.json
regenerated.
- Regression tests: all 10 name-guard paths + the duplicate-items path (19 cases).
- Docs: decisions.md entry + api-conventions.md §3b null-safe-validation bullet.
fixes#172
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold-fork + Codex review of PR #326:
- ApiRouteVersioningTests: iterate IRouteTemplateProvider (covers a
template-less [HttpGet] paired with an action-level [Route]) and assert
any non-/api route against an explicit KnownNonApiRoutes allowlist
instead of silently skipping — an accidental absolute non-/api route
(which would also escape ApiAuthorizationFilter's /api-scoped gate) now
fails the test.
- ApiVersionRewriteMiddleware: root the deprecation Link at Request.PathBase
so it stays correct under ETV_BASE_URL (</etv/docs>, not host-root </docs>).
refs #286#197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Version every /api route to /api/v1 (251 controller routes + ~24 Location
headers + the scanner callback URL + the Startup request-log literal),
uniform across the machine API, auth, scanner and scripted-build surfaces.
Add ApiVersionRewriteMiddleware: a legacy unversioned /api/* request is
rewritten (NOT redirected) to /api/v1/* in-pipeline — method, body, auth
headers and query survive — carrying RFC 8594 Deprecation/Sunset headers,
so curl / the future MCP server / bookmarks keep working. An already-
versioned path passes through; a future /api/v2 is never forced to v1.
Standardize the route convention (leading-slash absolute route per method,
no class-[Route] — except the two Scanner/Scripted controllers whose ~all
actions share a parametrized {id} prefix), enforced by ApiRouteVersioningTests
(^/api/v\d+/ over the whole Controllers.Api surface; browser-nav
/auth/oidc/login is out of scope).
Regenerate v1.json (160 paths, all /api/v1)/endpoint-index/v1.d.ts; sweep 945
SPA request literals + the test mocks (regex + positional URL parsers). /api/v1
is additive-only after freeze; the legacy-rewrite shim sunsets in ~2 releases
(owner decision) with removal tracked as a Phase-3 follow-up.
Docs: decisions.md 2026-07-13, api-conventions §1/§9, rest-api/spa-conventions/
blazor-route-parity/e2e-local/domain-model.
fixes#286
refs #197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold-review follow-ups (both non-blocking):
- Low: the TopBar "Add Trakt List" button was a silent no-op on the
/app/trakt-lists/{id} detail sub-route (setAddOpen state isn't rendered by
the editor branch, and the screen is keyed by pathname so the state
wouldn't survive a navigate). Guard on editingId: route back to the list
from the detail view, open the dialog from the list.
- Nit: the invariant test only spot-checked 2 screens. Replaced with a
data-driven it.each over the 4 URL-navigating create screens (channels,
filler, ffmpeg, watermarks) asserting each banner actually navigates — a
typo'd route id now fails red. Docs wording corrected to match.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The shell TopBar rendered a primary-action button (Plus icon) for every
screen, but only SchedulesScreen subscribed to its ctv:primary-action event
— so every other screen's button was dead (a labelled no-op, or a bare "+"
for the ~10 routes whose primaryAction was '').
Resolution (issue #238): the Plus-icon button is a "create new item"
affordance. Keep + wire it only on the 8 list screens with a single create
flow (channels, schedules, multi/rerun collections, trakt lists, filler
presets, ffmpeg profiles, watermarks) via a shared usePrimaryAction hook
(web/src/primaryAction.ts); drop it (primaryAction: '') everywhere else —
where the action isn't a create (Save/Refresh/Play/Validate/Reset/Scan, all
of which have correct in-body controls), is ambiguous (collections tabs), a
silent no-op (builder, playlists), or misplaced (dashboard, libraries). The
TopBar now renders the button only when primaryAction is non-empty.
Also relabels the apiKey route's stale post-#295 "Save key"/description.
Docs: spa-conventions.md §10 + decisions.md 2026-07-12.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The fix-as-you-touch format gate requires every .cs this PR touches to fully conform to
.editorconfig, including pre-existing whitespace on lines the change didn't edit. dotnet format
(whitespace) applied to the 11 touched files; legacy files left untouched (no big-bang reformat).
Whitespace/layout only — no behavior change. Refs #265
Independent review fix commit (cold fork MERGEABLE-WITH-NITS + Codex BLOCKED, 2 Highs):
- Codex H1: a comma (0x2C) is a valid etagc and can appear INSIDE a quoted opaque-tag
("3,5" is ONE tag). The old Split(',') broke it into two malformed tokens → 400. Replaced
with a quote-aware position scanner that treats a comma as a separator only outside the
quotes; "3,5" is now one valid non-canonical tag → 412.
- Codex H2: RFC 7230 OWS is SP/HTAB only. string.Trim() also strips NBSP and other Unicode
whitespace, letting " * " masquerade as the "*" force-write escape. Trim only
(' ', '\t'); such input is now Malformed → 400.
- Fork nit: corrected the canonical-guard comment (interior-whitespace tags are rejected by
IsEtagc, not NumberStyles.None).
- CI Formatting gate: de-BOM the 8 touched legacy Application .cs (charset=utf-8, #311/#310).
- Tests: added comma-in-tag ("3,5", "x,y","3"), empty-element tolerance, NBSP-not-OWS,
trailing-junk, lowercase-weak, wildcard-in-list cases. Full ErsatzTV.Tests green (1556).
Refs #253#197
The shared optimistic-concurrency parser (ConcurrencyHeaders.ParseIfMatch) classified any
non-canonical/weak/list If-Match value as Malformed → 400. Per RFC 7232 §3.1 a syntactically
-valid entity-tag that simply doesn't strong-match must be 412; 400 is only for a genuine
grammar violation.
- Rewrite ParseIfMatch as a real RFC 7232 entity-tag/list parser: walks the comma-separated
1#entity-tag list, validates each [W/]DQUOTE *etagc DQUOTE member, and collects the strong
members whose opaque text is our canonical decimal. Weak / empty / non-canonical /
out-of-range tags are valid but contribute no version (→ empty set → 412); genuine grammar
violations (unquoted, SP-in-tag, unterminated, garbage) → 400.
- Reshape IfMatchCondition.ExpectedVersion : Option<int> → ExpectedVersions : Option<Seq<int>>
and VersionedAggregateExtensions.CheckVersion → set membership (any strong match proceeds;
empty set always 412). Threads through 10 replace/update commands + handlers + request
mappers + 9 controllers.
- No wire-contract change (400 + 412 already declared on every PUT; the field is header-derived
and internal — no DTO/route/response-type/OpenAPI change).
- Tests: ConcurrencyHeadersTests rewritten for the new classification (lists, weak, empty,
non-canonical → Version/empty-set; grammar violations → Malformed) + new
VersionedAggregateExtensionsTests for CheckVersion membership/empty-set/force-write.
- Docs: api-conventions.md §7a rewritten; decisions.md entry appended.
Refs #253#197fixes#265
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Addresses the cold-review LOW: the guard hashed only the committed source
web/index.html, so a hypothetical future Vite transform of the inline script
would leave the test green while the deployed CSP silently broke the SPA. It now
hashes the built wwwroot/app/index.html when it exists (the exact bytes the
browser hashes — full coverage on any local build), falling back to the source
only in a fresh CI checkout where the built artifact is gitignored/absent. Vite
copies the inline script verbatim today, so the two agree. Doc comments +
decisions.md synced. Test-and-docs only; no production code change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Completes the CSP the #279 baseline-headers middleware deferred. Surfaced by
the #314 ZAP baseline (missing CSP/Permissions-Policy WARNs); a #197 exit item.
SecurityHeadersMiddleware now also sets Permissions-Policy (deny-all for
camera/mic/geolocation/payment/usb) and an enforcing Content-Security-Policy on
every response EXCEPT /docs (Scalar) and /openapi — those rely on inline
bootstrap scripts/styles a strict policy would break (baseline headers still
apply; hardening that admin surface is a #197 follow-up).
CSP: default-src 'self'; script-src 'self' + the hash of the SPA's inline
theme-bootstrap script (no 'unsafe-inline'/'unsafe-eval'); style-src adds
'unsafe-inline' (React inline styles) + fonts.googleapis.com; font-src adds
fonts.gstatic.com; img-src adds data:/blob:; object-src 'none'; base-uri 'self';
frame-ancestors 'none'; form-action 'self'.
The Google Fonts allowance (the SPA CSS @imports the Geist web font) was caught
by live-E2E, which the static recon missed. A guard test hashes the committed
web/index.html inline script and fails if it drifts from the middleware constant
(the built wwwroot/app is gitignored/absent in CI, and Vite copies the inline
script verbatim). Verified: full test pass, live-E2E (SPA renders clean, zero
CSP violations), curl (CSP present on /app + /api, absent on /docs + /openapi).
Docs: docs/decisions.md entry.
fixes#319
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The new api-docs gate (#303 H4/H5) runs update-openapi.sh after only `dotnet
restore`. The script's `dotnet build -t:GenerateOpenApiDocuments` does not compile
the project (OpenApiGenerateDocumentsOnBuild=false), so in a clean tree getdocument
fails with 'ErsatzTV.deps.json does not exist' (exit 129). This PR is the first to
change the /api surface and thus the first to exercise the gate's regen path,
exposing the latent bug. Add a full `dotnet build` before the doc-gen target so the
assembly + deps.json exist. Verified: clean-tree regen now succeeds with zero diff.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- GET /api/graphics-elements no longer side-effects; refresh moved to
POST /api/graphics-elements/refresh (204), closing a CSRF vector on a GET.
- PrepareTroubleshootingPlaybackHandler now returns a typed LockedError from
both atomic lock-acquire failures; ApiResults.ToErrorResult maps it to 409
instead of falling through to 422, so a lock lost in the race between the
controller's pre-check and the handler's atomic acquire still reports 409.
- AuthController.MachineKey sets Cache-Control: no-store + Pragma: no-cache
on the 200 response carrying the master API key.
- Reworded the stale "subtitleId query parameter" endpoint description now
that playback/start takes a JSON body.
- Regenerated openapi/v1.json + docs/endpoint-index.md; docs/api-conventions.md
updated with the LockedError pattern (§3a) and the ToErrorResult table row.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Slice A updated OpenApiContractHonestyTests for the troubleshoot GET->POST route
change but missed the hardcoded [TestCase] rows in OpenApiErrorResponseContractTests
still referencing the removed GET/HEAD /api/troubleshoot/playback.m3u8 409 — the
lookup threw KeyNotFound on the regenerated spec. Point them at the new
POST /api/troubleshoot/playback/start (404/409/422, all ProblemDetails-documented).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The server returns MachineKeyResponse(string ApiKey) -> JSON { apiKey }, but the
hand-written SPA MachineKey type declared { key } and ApiKeyScreen read result.key,
which would be undefined at runtime (blank key + empty copy). Mocked unit tests
passed against the wrong shape. Align the type, the screen, and both test mocks to
the real { apiKey } contract.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the API-key SPA model with the session-cookie auth from PR1:
- client.ts: stop sending X-Api-Key; attach X-Csrf on mutating verbs (POST/PUT/
PATCH/DELETE); add suppressUnauthorizedSignal to skip the global 401 signal on
expected wrong-credentials 401s (login / change-password).
- api/auth.ts: hand-written wire types (AuthConfig/AuthSession/MachineKey — the
auth surface is IgnoreApi, deliberately not in generated types) + endpoint fns
(getAuthConfig/getAuthSession/login/setup/logout/changePassword/getMachineKey);
keep the notifyUnauthorized/subscribeUnauthorized 401 signal; add
clearLegacyStoredApiKey. Legacy get/set/clearStoredApiKey retained ONLY so the
still-shipping ApiKeyScreen (machine-key slice) compiles without a cross-slice
conflict — the client no longer reads them.
- AuthGate.tsx: boot gate wrapping <App/> outside the shell (mints no URL, deep
links survive login). checking -> setup | login | ready | error, with a safe
default AuthContext so App.test.tsx renders without a provider; config-fetch
failure lands on an explicit error+Retry, never a blank screen.
- LoginScreen / SetupScreen: shell-less centered cards; inline 401 / 409 handling;
SSO button + local-form gating from AuthConfig.
- UnauthorizedBanner: rewritten to prompt re-login (passive; consults the
unsaved-changes guard before flipping the gate).
- UserMenu: TopBar sign-out (guard -> logout -> signOut), mounted next to
ConnectMenu.
- main.tsx: wrap <AuthGate><App/></AuthGate> inside StrictMode.
Tests: client/auth/AuthGate/LoginScreen/SetupScreen/UnauthorizedBanner/UserMenu
(723 pass). Lint + build green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adversarial review (MERGEABLE-WITH-NITS) findings:
- Remove the broad FAIL_RE log-scan (matched benign ErsatzTV startup noise — library
scans against absent media mounts, EF connection retries — risking a false-FAIL that
blocks a good release). It was also redundant: a failed migration faults the
BackgroundService -> default StopHost -> container exit, which the early-exit check
already catches reliably (per the reviewer's own analysis). Migration failure is now
detected by early container exit + timeout + the post-boot serve probe.
- Validate --timeout is a positive integer (was: '--timeout abc' -> 0 -> instant false-FAIL).
- Log the resolved image id after (attempted) pull, so a pull-failure that rehearses a
stale local :latest is visible to the operator.
Re-validated live on bumblebee: :latest vs the 283MB prod-copy -> migrations clean, PASS,
image digest logged, no leftover temp dir/container. shellcheck + bash -n clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
CI's migrations job only proves a migration is well-formed against a fresh,
empty DB. It never exercises the migration — or ErsatzTV's startup data steps
(DatabaseMigratorService -> DbInitializer + PopulatePathHashes over the real
MediaFile table) — against the accumulated prod SQLite, so a migration green on
a fresh DB can still fail/corrupt on prod, found only mid-deploy.
scripts/migration-smoke.sh rehearses it on a THROWAWAY copy of the latest prod
backup: boots the new image against the copy, gates PASS on the "Done applying
database migrations" log line (the migrator is a BackgroundService running
concurrently with Kestrel, so HTTP readiness alone doesn't prove migrations
finished), FAILs on early container exit / migration exception / timeout / not
serving afterwards. Always operates on a copy, never the live DB; tears down its
container + temp dir (incl. the container's root-owned config files) on exit.
Validated live 2026-07-12: :latest vs a copy of the 283MB prod backup ->
migrations applied cleanly, app booted+served, temp dir removed.
Home split: this repo owns the script + docs; wiring it into the Komodo
pre-deploy step is server-management#589 (cross-repo). Docs: docs/ci-cd.md
(Migration integrity), docs/decisions.md (new entry, pure insertion).
fixes#315
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adversarial-review nit on the first commit: auto-granting the docs-only exemption
silently self-merges process-control PRs (.claude/.gitea/.husky — including the
gate hook itself) with no prompt and no review, bypassing human-in-the-loop for
exactly the files that control the gate. Restrict auto-grant to the genuinely-
satisfied (a+b+c) merge path; the docs/process exemption reverts to bare exit-0
passthrough (one normal prompt). Corrects this PR's own decisions.md entry
accordingly ([decisions-edit]: a not-yet-merged draft entry, not history).
8 pipe tests green (adds docs-only -> passthrough).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The H6/H10 merge-consent hook's satisfied path did a bare `exit 0`, which does
not auto-approve a PreToolUse tool call — it only declines to block, so control
fell through to the normal permission system and the raw MCP prompt still fired.
A ready-to-merge PR was therefore confirmed twice (conversationally + a redundant
mechanical prompt). Emit permissionDecision "allow" (new `grant` decision) on the
satisfied and docs-exempt paths so the derived state IS the consent; deny/ask
unchanged (fail-closed); non-merge methods keep the exit-0 passthrough.
Docs: CLAUDE.md, kickoff HARD CONSTRAINTS, docs/decisions.md (append-only, pure insert).
Verified: 7 pipe tests (satisfied->allow, unticked->deny, stale->deny, red-CI->deny,
no-verdict->ask, no-creds->ask, non-merge->passthrough).
fixes#317
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
scripts/issue-qualification-audit.sh: lists OPEN issues missing a `priority:` label
(the #237 ranking keys off priority:/gate labels, so an unlabeled issue is invisible
to it). "Fully qualified" = has a priority: label; gate-vs-backlog derives from the
review label / milestone, and a milestone is NOT required (backlog is unmilestoned).
Advisory (exit 1 if any unqualified); fail-open without Gitea creds. Wired into the
kickoff session-end protocol + a lore bullet. Tested live (flagged 2) + no-creds no-op.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- contributing.md §7: document fix-formatting-as-you-touch (normalize a legacy file you
edit, incl. BOM strip; no big-bang), enforced by the pre-commit hook + the new format CI job.
- kickoff lore: flip "merge main into PR branch" -> "rebase on origin/main" (H11 enforces it);
keep the regenerate-generated-artifacts-on-conflict nugget.
- decisions.md: new entry (+ TOC) for both decisions.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- H11: .husky/pre-push calls .claude/hooks/prepush-rebase-check.sh, which blocks a
push whose branch is behind origin/main (rebase first; do not merge main in — a
merge drags in files you didn't touch, e.g. legacy-BOM .cs, tripping the format
hook on code that isn't yours). Fail-open; escape ETV_SKIP_REBASE_CHECK=1.
- New blocking `format` CI job: dotnet format --verify-no-changes scoped to the
PR's changed .cs only (style + charset=utf-8/no-BOM), enforcing fix-as-you-touch
without a big-bang reformat of the ~2500 legacy BOM files. .cs-free PRs skip and
pass (always reports a status). Closes the "CI never checks charset" gap that let
#269 land 17 BOM files (#310).
Docs (contributing.md §7 / decisions.md / lore) follow in the next commit.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two of the four non-hook #303 process follow-ups (the two docs items; the
security scan and migration-on-prod-copy smoke are deferred to their own
sessions):
1. Codex-skip rubric (kickoff workflow lore): an independent review pass is
MANDATORY for diffs touching locks/concurrency, auth/security, API
write-path handlers, or migrations, or >~150 changed C# lines; skippable
only for a pure-SPA/docs leaf, and a skip must be stated + justified. Makes
self-exemption an auditable claim (the correlated-blindspot net).
2. Live-E2E is now a STATED REQUIREMENT for API write-path handler changes:
new "When live-E2E is required" section in docs/e2e-local.md + a decisions.md
entry, formalizing the #229 lore bullet. The seeding recipe was already in
e2e-local.md (added for #220), so the stale "recipe not yet in docs" lore
bullet is pruned to a pointer.
Docs-only.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Independent Codex review (reconciled by Fable against a MERGEABLE fork verdict) found
SaveChangesForcingVersion silently DROPPED a pending Version++ under a concurrent
versioned-write race: on DbUpdateConcurrencyException it adopted the DB's current
Version verbatim (original = current = dbVersion), so a bumping sibling committed at
dbVersion instead of dbVersion+1. Net: an editor holding the concurrent writer's ETag
was never invalidated by the sibling's change — the exact lost-update the #253/#269
contract exists to close, lost under the very condition the helper handles.
F1 fix (shared helper, corrects all 25 bumpers incl. the pre-existing Add*ToPlaylist /
schedule-item writers): rebase the pending delta on top of the stored token —
pendingDelta = current - original; original = dbVersion; current = dbVersion + pendingDelta
Bumpers (delta 1) advance to dbVersion+1; non-bumpers/deletes (delta 0, e.g.
ErasePlayoutHistory) still adopt the stored token unchanged, so RootWriterForceVersionTests
is unaffected. Idempotent across the bounded retry loop.
F3: the force-race tests now assert Version==3 (rebase), not just membership survival;
added the missing Playout force-race+rotate test. Negative-controlled: with the helper
fix reverted, both strengthened tests go red.
F2 (Medium, deferred → #308): two concurrent same-item Add*ToCollection can both pass the
membership check and the loser 500s on the composite-PK violation (DbUpdateException, which
the helper doesn't catch). Pre-existing and narrow (no corruption); doc claims softened to
name it. Filed #308.
Docs: api-conventions §7a + decisions.md prose corrected from "adopt the stored token" to
the rebase semantics.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Complete the #253 optimistic-concurrency contract's cross-editor ETag
rotation tail. The non-If-Match config siblings mutated editor-visible
state without bumping Version, so a concurrent editor of the same root
never invalidated. Now the Collection Add*/Remove handlers bump
Collection.Version, and UpdateCollection / UpdatePlayout / the three
ScheduleFile writers (which already force-wrote past a concurrent bump)
now bump too — all via SaveChangesForcingVersion (no If-Match → force
write, never 412/500).
No-op idempotence (Fable-caught trap): these gate reindex/BuildPlayout
fan-out on SaveChanges()>0, so an unconditional bump would fire spurious
rebuilds on an idempotent re-add / same-value re-submit. Each now
short-circuits a genuine no-op before the bump — Add handlers by an
explicit membership check (also fixing a latent duplicate-CollectionItem
insert), scalar writers by ChangeTracker.HasChanges().
Corrects #269's framing: the Add*ToCollection family is not
repository-mediated (IMediaCollectionRepository is read-only); each
handler writes via its own dbContext, so the scanner's separate
membership path is unaffected (a background scan does not rotate the
editor ETag).
Tests: CollectionEtagRotationTests + PlayoutScheduleFileEtagRotationTests
(rotation, no-op-without-bump-or-rebuild, force-write-past-concurrent-bump),
no-op guard proven non-vacuous by inverting the membership check.
Docs: api-conventions §7a + decisions.md. No new status codes / no
OpenAPI change (these endpoints take no If-Match, never 412).
The #265 RFC-7232 If-Match parser refinement is a separate PR.
fixes#269
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Final re-review returned SHIP-IT with one contrived, pre-existing residual: a
line starting `Review-verdict: BLOCKED …` that ALSO contains a second literal
`review-verdict: mergeable` substring later on the same line read as positive.
Anchor the is_pos check to line-start so only the line's OWN leading verdict
word counts. Safe-by-construction: anchoring a positive matcher can only REDUCE
the allow-surface, so it cannot introduce a false-open (the dangerous
direction); the 21-case regression confirms no false-deny (all still
allow/deny/ask as before) + a new B6 case for this exact vector. No further
review round needed — the change is monotonic. shellcheck clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Second adversarial re-review (of the fix commit itself — dogfooding H10) found
a remaining false-open: the `review-verdict:` marker was matched anywhere on a
line, so a comment merely QUOTING the positive template — an instruction
("please post: Review-verdict: MERGEABLE @ <head>"), a blocking comment
explaining how to clear itself, or the gate's OWN suggestion text echoed into a
comment — was classified as a real head verdict and self-approved the merge.
Fix: anchor the marker to line-start (`^[[:space:]]*review-verdict:`). A real
verdict line starts with the marker; quoted/instructional uses have text before
it. Also drops the dead `nosha` var (SC2034).
Finding 2 (a BLOCKED mis-anchored to an OLDER sha doesn't retract a
MERGEABLE@head) is deliberately NOT "fixed": staleness is symmetric — a
pre-fix BLOCKED@oldsha must not block forever after the fix changes the sha and
earns a fresh MERGEABLE@head. To retract, re-review head and post BLOCKED@head.
Documented in decisions.md.
Pipe-tested 21 cases (5 new: instructional-quote, self-reference reason-text,
line-start-in-multiline, leading-whitespace, blocking-quotes-template), all
deny/ask/allow correct. bash -n + shellcheck clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adversarial review found false-opens in the first cut:
- `grep -F "$short"` was an unanchored substring test: a MERGEABLE verdict for
a DIFFERENT/older commit was accepted whenever the head 7-prefix appeared
anywhere on the line (inside a longer sha, or an unrelated commit URL). Now
each verdict line's `@ <sha>` token is extracted and matched to head by git
short-sha prefix semantics (head begins with token, token >=7 chars).
- No retraction semantics: a later `BLOCKED @ head` didn't override an earlier
`MERGEABLE @ head`. Now a negative verdict on head wins -> deny.
- A 7-digit build number falsely tripped the "references an older commit" deny;
the `@`-anchored parse fixes it -> a marker with no @sha now asks, not denies.
Also documents the issue-comment scope (gate reads issues/{pr}/comments, not
Gitea formal-review bodies). Pipe-tested: 16 cases incl. 4 adversarial
false-open reproductions, all now deny/ask. bash -n + shellcheck clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Folds condition (c) into pretooluse-merge-consent.sh (H6): a PR merge is
allowed only when a `Review-verdict:` comment references the PR's CURRENT
head sha — proving the latest commit was reviewed, not a stale earlier diff
(mechanizes the ersatztv#242 "re-review the fix commit" lesson).
Graceful adoption mirrors H6's Done-when tiering:
- positive verdict @ head -> allow
- verdict @ older sha (stale) -> deny (#242 failure mode)
- head verdict negative -> deny
- marker with no sha / none yet -> ask
- comments unfetchable -> ask
Reuses H6's PR fetch, docs-only exemption, and Gitea-auth-from-env (one hook,
no detection drift — per the #303 methodology review). Pipe-tested 12 cases.
Docs: decisions.md (new H10 entry + TOC), CLAUDE.md Task Completion Protocol.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Merge pull request '#303 Wave 3 — H3 root-screenshot guard + H9 decisions.md append-only guard' (#305) from ci/303-wave3-guards into main
Part of #303. Adds the Husky/CI append-only guard for docs/decisions.md (with the [decisions-edit] escape, mark-and-keep supersession, and a 1800-line read-cost consolidation floor) and the root-screenshot pre-commit guard. Reviewed MERGEABLE; user-approved merge.
Run 553 hit the known EF-migration MySql-apply flake (concurrent-runner contention;
MEMORY ci-migration-job-mysql-flake). This PR touches only hooks/docs/yaml — the
identical pipeline passed fully green on the parent commit (run 552). Empty commit to
get a clean run; Gitea 1.24 has no job-rerun API endpoint.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Timothy's refinement: the between-releases consolidation floor triggers on the file's
READ COST — its line count, i.e. the context an agent burns reading the log — not entry
count. The decisions-guard CI job now emits a non-blocking ::warning:: once decisions.md
exceeds 1800 lines (the point past which it no longer fits one default 2000-line agent
Read). Documented in the decisions.md header, the H9 entry, and ci-cd.md (job description
+ release-checklist note). Touches committed H9 lines, hence [decisions-edit].
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adversarial review (PR #305, MERGEABLE) nits:
- H3: `grep -iE` so a root `Screenshot.PNG` is caught too (was lowercase-only).
- decisions-guard.sh: comment the trailing-newline assumption (dropping the final
newline would make git render the next append as a last-line modify -> false-block;
self-correcting via [decisions-edit], .editorconfig enforces the newline).
- docs: clarify CI is PR-wide (`range`) vs Husky per-commit (`staged`) — shared
detection logic, deliberately different granularity; local hook is the stricter gate.
Replaces the slightly-overstated "can't drift" wording. Touches the committed H9
decisions.md entry, hence the [decisions-edit] token.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
decisions.md header now documents the enforced append-only rule, the `[decisions-edit]`
escape, mark-and-keep supersession (banner on the reversed entry + `(superseded)` TOC
tag), and consolidation at each release. ci-cd.md documents H3/H9 hooks + the
decisions-guard job and adds a "consolidate decisions.md" step to the release checklist.
New decisions.md entry records the H9/H3 mechanization. The header edit modifies existing
lines, so this commit carries the [decisions-edit] token — the guard working as designed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
H3 (Husky pre-commit): reject a staged root-level *.png — belt-and-suspenders with
the .gitignore screenshot rule so `git add -f` still can't land a review artifact.
H9 (append-only decisions.md): new shared hook `.claude/hooks/decisions-guard.sh`,
wired into Husky commit-msg (staged mode) and a new blocking `decisions-guard` CI job
(range mode). Blocks any commit/PR that deletes or modifies an existing line of
docs/decisions.md — detected via `git diff --numstat` deleted-count, robust to markdown
`-` list markers — unless the message carries the `[decisions-edit]` token. Pure
insertions (a normal new entry) always pass. One implementation for local + CI so they
can't drift. Fail-open on any tooling trouble.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Wave 2 hook H6: derive merge-consent from state instead of memory. An
issue's ## Done-when checklist (issue body) becomes the machine-readable
source of truth for whether its PR may merge — the structural fix for the
queue-drift #303 tracks (status was living in append-only prose).
- pretooluse-merge-consent.sh (Claude PreToolUse on mcp__gitea__
pull_request_write): a merge is ALLOWED only when the PR's CI is green
AND every ## Done-when box on the linked issue (fixes #N) is ticked;
DENY on an unticked box / red CI; ASK (human prompt) when state isn't
derivable (no linked issue, no section, no creds, Gitea down). Docs/
process-only PRs exempt.
- .husky/pre-push -> prepush-donewhen.sh: fail-open backstop for a direct
`git push origin main`; blocks only on a positively-proven unticked box.
Gitea auth from env only (ETV_GITEA_BASICAUTH / ETV_GITEA_TOKEN,
ETV_GITEA_URL) — nothing committed; without creds the gate degrades to
today's manual confirmation, never a silent pass. Non-breaking rollout:
until issues adopt ## Done-when the merge hook simply asks.
Pipe-tested: non-merge->allow, no-creds->ask, docs-only->allow, checklist
parser (unit), linked-issue extraction, and a live end-to-end block path
(temp Done-when on #303 -> exit 1 -> restored). Docs: CLAUDE.md Task
Completion Protocol + decisions.md entry. Refs #303.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds a blocking `api-docs` CI job: when a PR diff touches the API surface
(ErsatzTV/Controllers/Api/** or ErsatzTV.Core/Api/**) it rebuilds the
generated artifacts from source — v1.json, v1.d.ts, endpoint-index.md —
and fails if any is stale in the diff. Mechanizes the "docs-update in the
same PR" rule for the API contract (docs-reminder stays a non-blocking
route-parity nudge).
Path-gated INSIDE the job (per-step `if:` on a detect output), not via a
top-level `if:`, so the check always reports a status on every PR and is
safe as a required check: API-free PRs skip the dotnet/node setup + regen
and pass trivially.
Verified the gate reproduces the committed baseline: a fresh build
regenerates v1.json byte-identical to HEAD (incl. all 244 auth
security/401 blocks). The only footgun is local — update-openapi.sh runs
dotnet-getdocument against the already-built assembly, so a stale bin/
emits a stale spec; api-conventions.md §5 now flags "build first". CI is
immune (fresh checkout has no bin/).
Docs: api-conventions.md §5 (two-place CI enforcement + stale-assembly
note), decisions.md (new entry). Refs #303.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fix-commit re-review (cold fork) caught that e383c253 tracked a machine-specific
absolute-path symlink `web/node_modules -> /Users/.../web/node_modules` (created
for the eslint pre-push hook). It slipped past `.gitignore` because the
`web/node_modules/` trailing-slash pattern matches only a directory, not a
symlink; a real node_modules dir (a copy) would have been ignored. Untrack it and
tighten the ignore to `web/node_modules` (matches symlink or dir) so it can't
recur.
Also tightens the §7a / decisions.md "they already catch" phrasing (LOW review
nit): only BuildPlayoutHandler catches; PlayoutTimeShifter is insulated by running
solely on the background worker, never the request path.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Brings in the #303 process/rigor hooks + docs. Only conflict-free overlap was
docs/decisions.md (main added the ## Index TOC); reconciled by adding the #295
auth entry to the index.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Wave 2 hook H7: never commit/merge inside a sibling worktree another
session created (burned us on #289 path-leak + the plumbing-merge
workaround). Ownership = a per-session .claude-worktree-owner marker:
- posttooluse-worktree-marker.sh stamps a worktree with session_id on
`git worktree add` (parses the <path> arg past -b/-B/--reason flags).
- pretooluse-worktree-guard.sh denies `git commit`/`git merge` whose
effective dir (resolves `git -C <p>` and leading `cd <p> &&`) is a
worktree whose marker names a DIFFERENT session. Fail-open: no marker,
unparsable, or own session -> allow. Main tree + pre-convention
worktrees are never marked, so unaffected.
Also completes Wave 1's rollout, which committed pretooluse-bash-guard.sh
but left .claude/settings.json and the agent-ram/nav-guard hooks
untracked (so nothing was actually wired). Adds the settings.json that
registers all five hooks (PreToolUse Bash x2, nav, Agent; PostToolUse
Bash) + the .gitignore worktree-marker line, screenshot-scratch rules,
and the decisions.md TOC left uncommitted last session.
All hooks pipe-tested (7 guard cases + 6 marker cases). Refs #303.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Independent adversarial review (cold fork + Codex) of the first cut converged
on one real miss and two boundaries to document.
- **ErasePlayoutHistoryHandler** (HIGH, both reviewers): modifies Playout ROOT
scalars (Seed/Anchor/OnDemandCheckpoint) *without* bumping Version, inside an
explicit transaction with no try/catch, so it 500s on a concurrent bump —
reachable via POST /api/playouts/{id}/erase-items-and-history. My first sweep
filtered on "Version-bumpers + deletes"; the true exposure surface is "any
handler leaving a versioned root Modified/Deleted", so this slipped through.
Now routes through SaveChangesForcingVersion (+ a non-vacuous through-handler
test that exercises the explicit-transaction path). Re-swept with the correct
filter: ErasePlayoutItems (AsNoTracking + ExecuteDelete children only) and
ResetAllPlayouts (read-only + enqueue) are NOT exposed.
- **Background build/time-shift Playout-scalar writers** (BuildPlayout via
PlayoutBuilder, PlayoutTimeShifter): token-guarded too, but intentionally left
on plain save — they already catch (build-failure, not 500), and force-writing
would persist output built from stale config (the concurrent config bump already
enqueues a rebuild). Documented as a deliberate boundary, not a gap.
- **Item-add index collision** under force-write: documented as an accepted
Phase-1 effect (non-corrupting, self-correcting; reload-recompute refinement
is a #197 candidate).
Also corrects the docs' "every Version bumper" framing to the true filter and the
test docstring's over-broad non-vacuity claim. Full ErsatzTV.Tests green (1483).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fix-commit re-review confirmed the 1st-round fixes resolved and caught a 2nd round:
- HIGH — env-seed vs. setup race: an attacker could claim admin in the startup
window before LocalAdminSeedService runs, and the seed's insert would then be
swallowed (attacker credential persists, defeating env recovery). Fixed
structurally: the setup-claim endpoint is CLOSED (409) whenever
Auth:LocalAdmin:Password is configured — the env seed owns the credential, so
there's no claim to race (also strengthens the setup-claim TOFU posture).
Config.setupRequired reflects it.
- LOW — a concurrent setup race-loser now returns 409 (not 422); ClaimLocalAdmin's
DbUpdateException catch re-checks existence and rethrows genuine/transient DB
errors instead of masking them as "already configured".
- MEDIUM (accepted, documented) — two simultaneous authenticated password changes
are a non-serializable lost-update; accepted for a single-admin system
(self-healing via re-login, implausible timing).
+3 AuthController tests (env-seed closes setup / setupRequired gating). Full
ErsatzTV.Tests green (1506); no generated drift. Docs updated.
Refs #295
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Live E2E found that replaying a pre-logout cookie still authenticated (200, not
401): SignOutAsync only clears the CLIENT cookie, but the stateless encrypted
cookie ticket stays valid server-side because its security stamp is unchanged —
a captured cookie was replayable after logout until ticket expiry.
Fix: logout now rotates the local-admin security stamp (RotateLocalAdminSecurityStamp),
so every outstanding local session (old stamp) fails OnValidatePrincipal on its
next request. For the single admin this is "log out everywhere". Gated on an
authenticated local session so an unauthenticated caller can't force-revoke the
admin. OIDC sessions (no stamp) are unaffected; SignOutAsync still clears the
client cookie for UX.
+2 handler tests (rotate-when-configured / no-op-when-unconfigured). Auth suite
green (21). Docs: decisions.md note updated.
Refs #295
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Independent review (cold fork = MERGEABLE-WITH-NITS; Codex = BLOCKED, caught
concurrency defects the fork missed). All actionable findings folded in:
- HIGH (Codex) atomic first-claim-wins: ClaimLocalAdmin now writes the three
credential rows in ONE transaction guarded by the unique ConfigElement.Key
index (lost race -> DbUpdateException -> 409), so concurrent claims can't
produce a mixed-state credential.
- HIGH (Codex) consistent login snapshot: VerifyLocalAdminLogin reads hash+stamp
in one query and drops rehash-on-verify, so a login racing a password change
can't capture a stamp newer than the hash it verified (concurrent change ->
old password fails, or the issued cookie carries the pre-change stamp ->
revoked next request).
- MEDIUM (Codex) env-seed migration race: LocalAdminSeedService is now a RunOnce
BackgroundService that awaits SystemStartup.WaitForDatabase (the migrator is a
BackgroundService; registration order didn't guarantee the schema) + try/catch.
- MEDIUM (fork M1) ForwardedHeaders: reverted the strict-opt-in flip — it would
regress /iptv M3U/XMLTV/HLS absolute-URL generation (Request.Scheme) behind a
proxy without KnownProxies. Kept #285 behavior; KnownProxies still recommended.
- LOW (Codex/fork) require X-CSRF on /api/auth/logout + /password (the
[SkipApiAuthorization] surface isn't covered by the filter's CSRF check;
closes forced-logout CSRF).
- ChangeLocalAdminPassword also writes hash+stamp atomically. Input length caps
on username/password.
Deferred with a tracked gate: MEDIUM (Codex) side-effecting [RequiresAuthentication]
GETs (troubleshoot playback/archive) aren't CSRF-covered -> #301, gates PR2
(latent in PR1: the SPA still uses the machine key).
Verify: full ErsatzTV.Tests green (1501); no OpenAPI/generated drift. Docs updated
(api-conventions §9, decisions.md).
Refs #295#301
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Activating #253's `Version` as an `IsConcurrencyToken` made EF guard every
UPDATE *and DELETE* of a versioned root with `WHERE Version=@orig`, so any
writer outside the If-Match contract that saves via plain `SaveChangesAsync`
throws an unhandled `DbUpdateConcurrencyException`->500 when a replace-all
editor bumps the row in its narrow load->save window (ordinary two-tab UI).
A completeness sweep (grep every `Version` bumper + every root delete, not
just the handlers PR3's close note named) found 17 exposed writers, all now
routed through `ConcurrencyExtensions.SaveChangesForcingVersion` (Phase-1
force-write: adopt the stored token and retry; rethrow only on genuine
row-deletion):
- 9 versioned-root delete handlers (a delete has no ETag to rotate -> force
only, no bump)
- UpdateProgramScheduleHandler (bumps then saved plainly - the ProgramSchedule
case PR3 only suspected; its post-commit query/enqueue also moved to
CancellationToken.None per section 7b)
- 7 item add/remove bumpers PR2 left on plain save:
Add/DeleteProgramScheduleItem + Add{Items,Movie,Show,Season,Episode}ToPlaylist
Force-write (not 412) is correct: these endpoints take no If-Match, so an
unconditional delete/edit should win. No API contract change (no new response
codes) -> no OpenAPI regen.
Still deferred to #197 (cross-editor ETag rotation only, not a 500): the
non-bumping config siblings + the scanner-shared Add*ToCollection family.
Tests: RootWriterForceVersionTests races a bump *through the handler* via a
pre-tracked context (non-vacuous - reverting a handler to plain save fails the
test, verified) for the Option-delete / Either-delete / bump+update shapes,
plus the genuine-conflict rethrow branch and an explicit negative control
proving the plain-save path throws. Full ErsatzTV.Tests green (1482).
Also strips a pre-existing UTF-8 BOM from the touched handlers to satisfy the
.editorconfig `charset=utf-8` rule the pre-commit format hook enforces.
Docs: api-conventions section 7a (fan-out completeness) + decisions.md entry.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The lore had grown into a changelog (~40 verbose bullets). Removed bullets
whose substance lives in docs/decisions.md or api-conventions §7a/b/c
(#251 post-commit None, #252/#259 reconcile, #253 PR3/PR4 concurrency, addTo
layer, DTO nullable facts, disabled={saving}) and dead Blazor-parity process
notes (parity verdicts, screen!=parity, unwired-endpoint gaps, #205/#206
docs-staleness). Condensed the survivors. Added a scope banner: this section
is standing workflow/orchestration lore only; engineering decisions go in
decisions.md, never here (they duplicate and drift).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Recent session pickups re-did finished work (thought closed issues were
still open). Root cause: DONE/OPEN status lived in append-only prose (arc
inline markers + each comment's 'Recommended next'), which lags real issue
state — worst across parallel sessions narrating each other's merges.
Kickoff prompt fixes:
- SOURCE OF TRUTH = live Gitea state, never prose; verify OPEN (issue AND
milestone) before claiming.
- Gate/frontier defined structurally (lowest-numbered open arc item), not
hardcoded to an issue number (an earlier pass re-planted the bug by
hardcoding '#197 cluster'/'#91b milestone CLOSED').
- Label-based ranking: arc order -> gate(review label/milestone) -> priority.
- Pick order across arc-frontier / priority-pickup / audit pools.
- Bound the comment read (~6 newest); MCP-down REST fallback; #237 canonical.
- New lore bullet capturing the drift root cause + structural cure.
Companion edits to tracker #237 body (prune arc status -> 'Done (history)'
section) made live; noted in a session comment there.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Implements the ratified #295 design (PR1, server-only, backward compatible). The
/api surface now accepts a valid X-Api-Key (machine) OR an authenticated session
(browser cookie, local login or OIDC), gated by the evolved ApiAuthorizationFilter
(renamed from ApiKeyAuthorizationFilter; same fail-closed EndpointRequiresKey
predicate). Machine/key behavior is byte-identical and the SPA keeps working via
its stored key — the SPA login flow lands in PR2.
- ApiAuthorizationFilter: key-first (CSRF-immune) then session; session-authed
mutations require the X-CSRF header (403 otherwise). Attributes renamed
[RequiresApiKey]->[RequiresAuthentication], [SkipApiKeyAuthorization]->[SkipApiAuthorization].
- Cookie scheme ctv-session always registered (Lax/SameAsRequest/14d sliding, 401 not
redirect for /api); OIDC handler revived when configured (profile scope, userinfo,
auth-method claim); UseAuthentication/UseAuthorization/UseRateLimiter revived in the
legacy MapWhen branch.
- Local admin = single credential in ConfigElement rows (username / PBKDF2 hash via
Microsoft.Extensions.Identity.Core / rotating security stamp) — NO DB migration.
Password change rotates the stamp; CookieSecurityStampValidator revokes stale local
sessions. Env-seed recovery (Auth:LocalAdmin:*) via LocalAdminSeedService.
- AuthController /api/auth/{config,session,setup,login,logout,password} + browser-nav
GET /auth/oidc/login; excluded from OpenAPI (machine-audience spec). Per-IP rate limit
on login/setup/password; dummy-hash verify (no user enumeration).
- ForwardedHeaders now strict opt-in: X-Forwarded-* ignored unless KnownProxies/Networks
configured (rate-limiter IP + cookie-Secure integrity). Deployment: operators behind a
proxy must set ForwardedHeaders:KnownProxies.
- Tests: session/CSRF filter cases + 17 Application/Auth handler tests; full ErsatzTV.Tests
green (1499). No OpenAPI/generated-artifact drift.
- Docs: api-conventions section 9 rewritten; decisions.md entry (supersedes #206 inert-OIDC note).
Refs #295#197#206#58
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold-review nit fixes on PR #298:
- useCollectionsScan.scan() now guards on the whole family being busy (active OR
any pending key of that family), not just the exact key — a sibling source of a
family with a scan in flight no longer fires a redundant (benign-409) POST.
- LibrariesScreen ExternalCollectionsSection disables every row of a family that
has a pending or active scan (derives pendingFamilies from pendingKeys), matching
Blazor's instant all-rows-disabled behavior instead of waiting a poll RTT.
- Rewrite the promote test to actually observe the optimistic-pending window via a
deferred POST (was only asserting the promoted end state), and add a test proving
a sibling-source click fires no second POST while the family is pending.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add GET /api/media-sources/collections-scan-status (MediaSourcesController →
GetCollectionsScanStatus handler) reporting which media-source families
(plex/jellyfin/emby) currently hold their external-collections scan lock,
reading IEntityLocker.Are{X}CollectionsLocked(). The lock is family-global
(no source id) and boolean (no percent), so the DTO carries just {family} and
returns only active families — the counterpart to GET /api/libraries/scan-status.
SPA: useCollectionsScan now polls this endpoint and reconciles optimistic
pending against the active-family set (seeding on mount so an in-progress scan
disables buttons immediately), using the same grace-tick helper as library
scans (now generic over the pending key type). Drops COLLECTIONS_PENDING_TIMEOUT_MS
— a long deep scan no longer re-enables the button early, and a fast scan no
longer wedges it disabled for the full timeout. A row shows Scanning when its
family is active or it has an in-grace optimistic pending key.
Tests: handler (3), controller route+delegation (2), SPA api fn + hook reconcile
(mount-seed / 202-promote / 409-keeps-disabled / 404-error). OpenAPI + TS types
regenerated. Docs: api-conventions §3b, blazor-route-parity §5, decisions.md.
Unblocks #91b (arc item 4): Libraries.razor's collections-scan affordance now
has full authoritative parity.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Mint ChannelDetailResponseModel (faithful detail DTO exposing the raw editable
field set the channel editor reads: raw FFmpegProfileId/WatermarkId/FallbackFillerId
ids, the mode enums, logo, playoutCount, id) and route GetById/Create/Update through
it, replacing the lean list ChannelResponseModel that resolved the profile to a name
and dropped the editable ids (a functional regression for draftFromChannel). The lean
ChannelResponseModel stays unchanged for GET /api/channels. webEncodedName dropped
(SPA never reads it). Logo is mirrored as a Core ChannelLogoResponseModel since the
Application ArtworkContentTypeModel can't be referenced from Core.
Repoint the hand-written SPA client aliases now that the VMs are gone from the schema:
Channel -> ChannelDetailResponseModel, MediaCollection/SmartCollection -> *ResponseModel,
ProgramSchedule -> ProgramScheduleResponseModel. Fix#288 honest-nullability test fallout
in search.test.ts (null -> [] for now-non-null id arrays). Include the already-on-disk
playouts.ts WithDayNames removal and regenerate v1.json + v1.d.ts + endpoint-index.md
(authoritative final regen; the reset endpoint's {channelNumber}->{id} re-key surfaces
in the generated docs and the OpenApi error-contract test).
Refs #288#197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
GetById/Create/Update return ChannelResponseModel via new GetChannelByIdForApi
read-side query; POST /api/channels/{id:int}/playout/reset (new
GetPlayoutIdByChannelId; by-number kept for HlsSessionWorker broadcast).
Refs #288#197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ApiKey security scheme + per-op security/401 via shared EndpointRequiresKey
predicate (no drift from enforcement); synthesized stable operationIds;
400 ValidationProblemDetails on binding ops; DayOfWeek as string enum.
Refs #287#197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
24 #nullable enable flips across ErsatzTV.Core/Api; Collection/Schedule/
SmartCollection/Resolution VMs wrapped in ResponseModels (Version now
header-only, SPA-verified); pageNum threaded into GET /api/search.
Refs #288#197
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Root cause (diagnosed from run logs 513/515/516): the EF migration-integrity
job's "MySql apply all migrations to a fresh DB" step flakes when two migration
jobs land on the SAME runner host at once — each `services: mysql:8.4` container
starves the other, so the 787-migration replay either exceeds MySqlConnector's
30s default command timeout ("Command Timeout expired", run 513 on ci-runner) or
has its connection dropped mid-replay ("MySqlEndOfStreamException", run 516 on
bumblebee-runner). It's pure infra contention: `has-pending-model-changes` (the
model check) passes both providers, and the identical tree passes on a quieter
host (run 515). Both runners have both passed and failed — not one bad runner.
Fix (runner-agnostic, repo-owned workflow only — no runner-host change needed):
- Raise `DefaultCommandTimeout` to 300s in the MySql connection string.
- Wrap the apply in a 3× retry that resumes from `__EFMigrationsHistory` (EF
commits each migration in its own transaction, so an interrupted one rolls back
and the retry continues). A real migration failure fails on every attempt, so
the retry can't mask a genuine problem.
Docs: ci-cd.md migration-integrity section documents the contention + retry.
Refs #13#236
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Run 516's MySQL "apply all migrations to a fresh DB" died with
MySqlEndOfStreamException (incomplete response) — the shared MySQL service
container under concurrent-run load, same class as run 513's Command Timeout.
Model-drift (has-pending-model-changes) passed for BOTH providers, so the diff
is model-clean; another branch's run (515) passed the identical job. Contention
has cleared; re-triggering. Tree unchanged from b6f12f7e.
Refs #283
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Run 513's "EF migration integrity" job failed with "The Command Timeout
expired" applying all migrations to a fresh MySQL under concurrent-run
contention. The tree is unchanged from b6f12f7e (Build & test green there and
on the pre-clamp cf834d8b; the change touches no EF/model/migration code).
Empty commit to get a clean run.
Refs #283
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Re-review of the fix commit (MERGEABLE-WITH-NITS) noted the headline L3 rethrow branch
itself had no test. Add a reader seam (internal ResolveKey Func overload) and two tests:
unreadable existing file throws + does not overwrite; a delete race between File.Exists
and the read falls back to generate rather than failing boot.
Refs #197#280
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold adversarial review of PR #292 = MERGEABLE-WITH-NITS (no BLOCKER/HIGH). Addresses:
- M1: ApiKeyProvider's never-empty invariant was untested. Add ApiKeyProviderTests
covering WriteKey precedence, load-existing, empty-file regenerate, generate+persist,
0600 mode, and still-usable-key-when-persist-fails. ResolveKey extracted to an
internal seam taking the key path (InternalsVisibleTo ErsatzTV.Tests).
- L2: write-then-chmod race — the key was briefly world-readable. Persist now creates
the file 0600 atomically via FileStreamOptions.UnixCreateMode (then re-asserts).
- L3: a transient read error on an EXISTING key file silently regenerated + clobbered
it (invalidating every client key). ResolveKey now rethrows on an unreadable existing
file (fail loud) and only regenerates when the file is absent or empty.
L4 (LocalhostOnly XFF-spoof under default trust-all) and N5 (length oracle on a
fixed-width key) accepted as documented/cosmetic.
Refs #197#280
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold-review LOW (defense-in-depth): the serve path derived the Content-Type
from the stored file via Winista but only defaulted application/octet-stream on
a NULL sniff. A cache file whose bytes are HTML — a legacy entry poisoned before
the upload-sniff landed, or a hypothetical image/script polyglot — could still be
sniffed as text/html and served renderable (nosniff does not stop an explicitly
declared text/html). Clamp the sniffed type to ImageContentTypes.IsAccepted,
serving application/octet-stream for anything else, so the serve path can never
emit a renderable non-image type regardless of what bytes are on disk.
Refs #283
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
S4 stored-XSS + S9 upload-size DoS from the #197 cold API review.
The artwork path trusted client-supplied content types at both ends: upload
validated only the declared multipart Content-Type (never decoded the bytes),
and serving reflected a client `?contentType=` straight into the response
Content-Type on unauthenticated GET sinks (/iptv/logos, /artwork/watermarks).
Chain: upload <script> bytes as image/png -> GET ...?contentType=text/html
serves them as HTML in-origin. nosniff (#279) does not help because the server
explicitly declares text/html.
- Upload: derive the content type from the bytes via SkiaSharp SKCodec
(header-only, no decode -> no decompression-bomb path); reject non-images 422.
New ErsatzTV.Core/Images/ImageContentTypes as the single allow-list source.
Dropped the untrusted declared Content-Type from the UploadArtwork command.
- Serve: removed the ?contentType= reflection structurally -- dropped ContentType
from GetCachedImagePath and the [FromQuery] binding on GetImage/GetWatermark;
the handler always sniffs the file, defaulting application/octet-stream.
ArtworkContentTypeModel.UrlWithContentType is now the bare path; SPA previews
no longer append the query.
- Defense-in-depth: channel-logo / watermark {path, contentType} DTOs run through
ArtworkContentTypeModel.Sanitized(), blanking non-allow-listed types on write.
- S9: Kestrel MaxRequestBodySize from ETV_MAXIMUM_UPLOAD_MB rejects oversized
bodies during read (controller file.Length check kept as friendly-error backstop).
Both serve sinks are IgnoreApi, so no OpenAPI change. Tests: byte-sniff accept/
reject, Sanitized() allow-list, Location no longer carries ?contentType=.
Docs: api-conventions §4a + decisions.md 2026-07-12.
Refs #283#197#66
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Backend of #197 Bundle A (auth posture). Owner decisions: single API key;
Api:RequireKeyForReads defaults true (whole /api surface gated; /iptv streaming
+ guide unaffected — outside the filter's /api scope).
- #280 S1: writes are fail-closed. New IApiKeyProvider resolves the key once
(Api:WriteKey config, else persisted /config/api.key, else a generated 256-bit
key written 0600). The empty-key open branch is gone; there is no open mode.
- #282 S3/S5: reads under /api require the key when Api:RequireKeyForReads (default
true) or the endpoint carries the new [RequiresApiKey]. Applied [RequiresApiKey]
to Troubleshoot/Logs/Settings/Maintenance so the sensitive tier stays gated even
if reads are opened. OPTIONS preflight is exempt.
- #281 S2: delete SortController (dead Blazor SortableJS residue; SPA uses PUT
/api/collections/{id}/custom-order) and AccountController (dead OIDC logout) —
both non-/api persistent surfaces that bypassed the key.
- #284 S6: replace CORS AllowAll with an opt-in exact-origin allowlist
(Api:CorsAllowedOrigins; permits X-Api-Key/If-Match, exposes ETag). Default is
no cross-origin (SPA is same-origin).
- #285 S7/S10: gc GET->POST (spec regenerated); ForwardedHeaders trust configurable
via ForwardedHeaders:KnownProxies/KnownNetworks (warns when unrestricted);
ScannerController gains [LocalhostOnly] (scanner always calls back over localhost).
Filter unit tests rewritten for fail-closed + read-gating + tier + OPTIONS;
ApiControllerSecurityTests assert the sensitive tier + scanner-loopback reflectively.
search/all-items paging deferred (SPA add-all coupling) — exposure closed by read-gating.
Refs #197#280#281#282#284#285
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bundle A SPA slice: the /api surface is now gated behind X-Api-Key on
every request (reads too, RequireKeyForReads defaults true), so a wrong/
missing key 401s everything.
- #282: send X-Api-Key on ALL requests when a key is stored, not only
mutations (removed the mutatingMethods split in api/client.ts).
- #280: new keyless API Key screen (/app/api-key, System nav) that reads/
writes only localStorage via auth.ts and never calls /api, so it works
on a fresh install where every read 401s. Masked key state, Save/Clear,
points at server-generated /config/api.key.
- 401 UX: client emits one app-wide unauthorized signal (auth.ts
notify/subscribeUnauthorized); a shell-level UnauthorizedBanner points
the user at the API Key screen. DRY, no per-screen 401 branches.
- Tests: inverted the GET header assertion (key now sent on reads), added
no-key and 401-signal client tests, auth signal tests, and screen +
banner tests. spa-conventions.md §5e documents the new seams.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Follow-up to PR #279 — the adversarial diff review flagged that adding
baseline security headers to every response is an operational-behavior
decision worth a decisions.md entry. Records the SecurityHeadersMiddleware
placement + the deliberate CSP/HSTS deferral to the #197 posture design,
plus the constant-time key compare and playout paging clamps.
Refs #197.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Posture-independent safe hardening from the #197 cold API security review
(the clear-cut fixes that don't depend on the fail-closed/CORS/versioning
posture design, which is tracked separately):
- ApiKeyAuthorizationFilter: compare X-Api-Key with
CryptographicOperations.FixedTimeEquals instead of ordinal string.Equals
(removes the response-timing oracle on the write key). [S10]
- PlayoutController: clamp pageNum/pageSize on GET /api/playouts and
/api/playouts/{id}/items to Math.Clamp(_, 1, 100), matching the documented
api-conventions §1 convention every other paged endpoint already follows —
these two were passing the raw value straight to EF Take(). [S8]
- SecurityHeadersMiddleware: emit X-Content-Type-Options: nosniff,
X-Frame-Options: DENY, Referrer-Policy: strict-origin-when-cross-origin on
every response (nosniff backstops the artwork content-type MIME-sniffing
risk). CSP/HSTS deferred to the #197 posture design (CSP needs SPA
validation; HSTS is proxy/TLS-owned). [S10]
Refs #197.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Timothy reversed the version-pin decision: prod's media-servers compose now
follows the floating :prod tag, redeployed by Komodo Global Auto Update. The
bump-prod-compose job (#275) rewrote a :<version> pin, which would flip :prod ->
:26.8.0 on the next release — remove it. docs/ci-cd.md reconciled to the :prod
model (+ flags the open caveat: verify Global Auto Update runs the #553
pre-deploy backup, else releases deploy without a backup).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Review-caught coverage gap: the existing Subtype_Change test runs the POSITIONAL
path (id-less payload). Add a handler-level test for the id-mode branch — one
id-matched item changes subtype (One->Duration: delete+insert, new id) while a
sibling id-matched item keeps its subtype (Multiple: in place, id + fill-group
state preserved) in the same payload. Proves the delete pass + match-pass Remove/Add
don't double-handle and the survivor's state is retained.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the id-based reconcile tests to ReplaceProgramScheduleItemsReconcileTests:
reorder moves state with the logical item (the non-vacuous core — proven to fail
under forced-positional), insert-in-middle, delete-unreferenced, unknown-id→422,
duplicate-id→422, and stale-version+unknown-id→412 (412 precedes 422, §7c). The
GET→map→PUT lossless round-trip now round-trips r.Id so it exercises id-mode.
Threads the new int? Id through all command/wire construction sites in tests.
Docs: api-conventions §7c (stable child identity + the deliberate #2-#5 positional
asymmetry) and a decisions.md entry.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The backend (already on this branch) added an optional int? Id to
ScheduleItemRequest so the server reconciles PUT /api/schedules/{id}/items
rows by identity instead of by array position. The SPA previously
discarded the server id on load (only a client-local _key survived) and
never sent one back, so a reorder could misattribute fill-group/shuffle
state onto the wrong persisted row.
- itemRules.ts: fromResponse now captures the response item's id onto
the draft; normalizeForSave emits it back unchanged. newDraftItem and
copyDraftItem explicitly set id: null (a brand-new/copied row was
never persisted under an id, and copyDraftItem must not duplicate the
source's id onto a second row).
- scheduleItem.ts (Add-to-schedule dialog, POST path): id: null for the
same reason — it always creates a new row.
- SchedulesScreen.tsx save(): the PUT-response re-seed already existed
(fromResponse over the response array) but now carries ids through.
This matters because a subtype/playout-mode switch can be a
delete+insert server-side, so the response id for that row can differ
from what was submitted — a second save must use the *response's* id
or the server 422s it as unknown. Added a comment documenting this.
- schedules.ts: replaced the stale "server reuses same-typed rows by
position" comment with the current id-based reconcile contract.
- Added/updated tests in itemRules.test.ts, SchedulesScreen.test.tsx,
and AddToScheduleDialog.test.tsx covering id round-tripping, the
null-id-for-new/copied-item cases, and a second-save-reuses-the-
response-id regression test.
Verified: npm run lint, tsc -b --noEmit, npm run build, and npm test
(680/680) all pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The auto-pin-to-prod job designed on the unmerged `ci/auto-bump-prod-compose`
branch (3d6ac883) never landed on main — so v* releases (v26.5.0, v26.6.0) did
NOT auto-bump the server-management compose pin (it sat at 26.5.0). The docs
(homelab-docs Docker/ErsatzTV.md, ci-cd.md) described the auto-bump as if live.
Restore the job verbatim (its credentials already exist: the `ersatztv-ci-deploy`
write deploy key, id 5, on server-management + the SERVERMGMT_DEPLOY_KEY secret
here). On a v* tag, after the test-gated image builds, it rewrites the pinned
`ersatztv:<version>` tag in docker/bumblebee/stacks/media-servers/compose.yaml
and pushes to server-management `master` → the Gitea->Komodo webhook redeploys
prod with a pre-deploy backup. Idempotent (no-op if already pinned).
docs/ci-cd.md updated to match (release procedure + the stale ":prod pin" claim,
which was actually an immutable :<version> pin since 2026-07-07).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add optional `int? Id` to ScheduleItemRequest/ReplaceProgramScheduleItem so
a client can round-trip each existing item's server id. When ids are present,
ReplaceProgramScheduleItemsHandler reconciles by id (not array position), so an
item's persisted fill-group/shuffle state (PlayoutScheduleItemFillGroupIndex,
FK OnDelete Cascade) follows the logical item across reorders/inserts instead of
being inherited by whatever previously occupied its new slot (#259, split from
#252/#253). A fully id-less payload keeps the verbatim positional fallback.
Guards (inside PersistItems, after CheckVersion so 412 precedes 422): duplicate
id -> 422; id not in this schedule -> 422 (a stale id under Phase-1 force-write is
a live lost-update signal, not a new item). Index stays array-position derived.
Regenerated v1.json + TS client.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Codex-review nits: drop the now-inert "MudBlazor" Serilog level override in
appsettings.json (package removed) and reword the PlaylistController comment
that referenced the deleted Blazor MultiSelectBase.AddItemsToPlaylist path. No
behavior change.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cold-review Low-1: ErsatzTV/Extensions/NavigationManagerExtensions.cs survived
the removal — the last non-deleted .cs still importing
Microsoft.AspNetCore.Components/JSInterop and calling the deleted
blazorHelpers.scrollToFragment JS. Fully unreferenced (compiled only via the
shared framework). Removing it completes the Blazor deletion.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The ChicoryTV React SPA (web/, served at /app) now has full parity for every
route the Blazor UI served, so the legacy Blazor Server / MudBlazor UI is
deleted. This is the milestone-capping removal of #91 phase (b).
Deleted: ErsatzTV/Pages/**, Shared/**, ViewModels/** (39 edit VMs),
Validators/** (10 edit-VM validators), App.razor, _Imports.razor,
Locals/{Shared,Pages}/** (Blazor loc resx; Locals/Resources.* kept),
wwwroot/css + wwwroot/lib, libman.json, and the orphaned MultiSelectBaseTests.
Startup.cs (surgical, not wholesale): removed AddRazorPages/AuthorizeFolder,
AddServerSideBlazor, AddMudServices, AddSortable, AddCourier, the HtmlSanitizer
registration, the Blazor-attached OIDC UseAuthentication/UseAuthorization
middleware (per the #206 auth-posture sign-off), MapBlazorHub, and
MapFallbackToPage("/_Host"). Renamed the branch blazor->legacy; it still
co-hosts MapControllers, /docs (Scalar), dev MapOpenApi and the redirect
middleware. Replaced the _Host fallback with a catch-all (MapFallback ->
302 /app) that excludes /api|/artwork|/docs|/openapi (genuine 404) per #204.
Kept all OIDC/JWT/API-key service wiring (inert unless configured; real auth
is #197), ConditionalIptvAuthorizeFilter, ApiKeyAuthorizationFilter.
Pruned 9 now-unused packages (all verified zero remaining consumers) from
Directory.Packages.props + ErsatzTV.csproj: MudBlazor, Heron.MudCalendar,
Blazored.FluentValidation, BlazorSortable, MediatR.Courier.DependencyInjection,
Markdig, HtmlSanitizer, Chronic.Core, NaturalSort.Extension. Also removed the
now-dead #25 razor-Sonar NoWarn.
LegacyUiRedirects: added the 14 /media/sources/* -> /app/libraries/* redirects
(SPA screens landed in #202) and lifted the #204-era /media/sources prefix ban.
Tests: Release build clean; full solution suite green. Updated Startup
source-text tests + added regression coverage that Blazor wiring is gone, the
catch-all is wired, and all 14 media-sources routes redirect.
Docs: blazor-route-parity.md (phase b COMPLETE), decisions.md (removal entry),
CLAUDE.md, contributing.md, README.md all updated in this PR.
Rollback: tag blazor-final is cut on pre-merge main as the first merge action.
Part of #91.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-11 21:01:04 +02:00
711 changed files with 26569 additions and 40781 deletions
jq -n --arg f "$free"'{hookSpecificOutput:{hookEventName:"PreToolUse",permissionDecision:"deny",permissionDecisionReason:("Free RAM \($f)% (<10%): do NOT spawn more agents — the historic crash was RAM starvation from an 8-9-way fan-out. Wait for memory_pressure -Q to recover, then retry.")}}'
elif["$free" -lt 20];then
jq -n --arg f "$free"'{hookSpecificOutput:{hookEventName:"PreToolUse",permissionDecision:"ask",permissionDecisionReason:("Free RAM \($f)% (<20%): near the fan-out ceiling. Confirm before adding another build/implementer agent (read-only recon agents are cheap).")}}'
jq -n '{hookSpecificOutput:{hookEventName:"PreToolUse",permissionDecision:"deny",permissionDecisionReason:"Blocked: ETV_UPDATE_GOLDENS regenerates golden-test baselines — HARD RULE (docs/handoffs lore); never set it in a session. Update a golden deliberately and reviewed, not via a guarded run."}}'
[ -n "$owner"]&&[ -n "$repo"]&&[ -n "$pr"]|| decide ask "H6 merge gate: could not read owner/repo/pull_number from the merge call; confirm manually that CI is green and the issue's Done-when boxes are ticked."
decide ask "H6 merge gate: no Gitea credentials in env (ETV_GITEA_TOKEN or ETV_GITEA_BASICAUTH), so CI/Done-when state can't be verified. Confirm manually that CI is green and the linked issue's Done-when boxes are all ticked, then approve."
fi
prjson=$(gq "repos/$owner/$repo/pulls/$pr")
[ -n "$prjson"]|| decide ask "H6 merge gate: could not fetch PR #$pr from Gitea (unreachable or auth rejected). Verify CI-green + Done-when manually before merging."
[ -n "$issues"]|| decide ask "H6 merge gate: PR #$pr has no linked issue (no 'fixes #N' / 'closes #N' in its body), so there is no Done-when checklist to derive consent from. Confirm the work is complete + reviewed, then approve."
# --- (b) Done-when checkboxes: every linked issue must have an all-ticked section. ---
"") decide ask "H6 merge gate: could not read CI status for PR #$pr ($sha). Verify CI is green before merging.";;
*) decide deny "H6 merge gate: BLOCKED — PR #$pr CI status is '$state', not 'success'. Wait for a green build (or pass merge_when_checks_succeed to let Gitea gate it) before merging.";;
esac
fi
# --- (c) Review-verdict freshness (ersatztv#303 H10): a review-verdict comment must reference the
# CURRENT head sha, so the latest commit is proven-reviewed (ersatztv#242: re-review the fix
# commit, not just the initial diff). Graceful adoption mirrors (b): a verdict comment that
# references head must be positive -> allow; one that exists only for an OLDER commit -> deny
# (the stale-review failure mode); NO verdict comment at all -> ask (convention not yet used).
[ -n "$sha"]|| decide ask "H10 merge gate: could not resolve PR #$pr head sha to verify a review verdict. Confirm the review covered the latest commit before merging."
decide ask "H10 merge gate: could not fetch PR #$pr comments to verify a head-referencing review verdict ($short). Confirm the adversarial/Codex review covered the latest commit before merging."
fi
# Verdict lines across all comment bodies: a real verdict line STARTS with the marker (after optional
# leading whitespace). Anchoring to line-start is deliberate — it rejects a comment that merely QUOTES
decide ask "H10 merge gate: no 'Review-verdict:' comment found on PR #$pr referencing head $short. Post the adversarial/Codex verdict (e.g. 'Review-verdict: MERGEABLE @ $short'), or confirm the review covered the latest commit and approve."
fi
# Classify each verdict line by the sha it references (its "@ <sha>" field) and its verdict word.
# A line references the CURRENT head iff head BEGINS WITH that sha token AND the token is >=7 chars
# (git short-sha prefix semantics) — NOT a loose substring test: an older sha that merely contains
# the head prefix, or the head prefix appearing in an unrelated URL on the line, must NOT count
# (adversarial false-opens). The verdict token must sit right after the marker on the same line.
head_pos=0;head_neg=0;stale=0
whileIFS=read -r line;do
[ -n "$line"]||continue
# The sha the line references: the hex token in its "@ <sha>" field (>=7 chars), lowercased.
ref=$(printf'%s'"$line"| grep -ioE '@[[:space:]]*[0-9a-f]{7,40}'| head -1 \
# A negative verdict on head wins over a positive one (a later BLOCKED retracts an earlier MERGEABLE
# on the SAME head; and if the head were fixed the sha would change, so this can't wrongly block).
if["$head_neg"=1];then
decide deny "H10 merge gate: BLOCKED — a review verdict for the current head ($short) is negative (BLOCKED/NOT-MERGEABLE). Resolve the findings and post a fresh 'Review-verdict: MERGEABLE @ $short' before merging PR #$pr."
fi
if["$head_pos"=1];then
# (a) CI green + (b) all Done-when ticked + (c) positive verdict @ current head -> SATISFIED. Auto-grant.
decide grant "H6/H10 merge gate: satisfied — CI green, all Done-when boxes ticked, and a positive Review-verdict references the current head ($short). Auto-granted (no separate confirmation needed)."
fi
if["$stale"=1];then
decide deny "H10 merge gate: BLOCKED — a review-verdict comment references an older commit, not the current head ($short). The latest commit(s) are unreviewed (ersatztv#242: re-review the fix commit, not just the initial diff). Re-review the head and post 'Review-verdict: MERGEABLE @ $short'."
fi
# Marker(s) exist but reference no sha at all -> ask (don't mislabel as a stale older-commit review).
decide ask "H10 merge gate: a 'Review-verdict:' comment on PR #$pr references no commit sha. Post one referencing the current head ($short) — e.g. 'Review-verdict: MERGEABLE @ $short' — or confirm the review covered the latest commit and approve."
# All derivable and satisfied -> auto-grant (defensive: the head_pos branch above already exits here).
decide grant "H6/H10 merge gate: satisfied — auto-granted."
jq -n '{hookSpecificOutput:{hookEventName:"PreToolUse",permissionDecision:"deny",permissionDecisionReason:"Blocked: do not open download/stream endpoints (/iptv, .m3u8, /artwork, playback.m3u8) in a browser tab — they stall the MCP session. curl them instead (docs/handoffs lore)."}}'
jq -n --arg o "$owner" --arg r "$root"'{hookSpecificOutput:{hookEventName:"PreToolUse",permissionDecision:"deny",permissionDecisionReason:("Blocked: worktree \($r) is owned by session \($o), not this one. Never commit/merge inside a sibling worktree another session created (#289 path-leak, plumbing-merge workaround). Commit from your own tree; if you genuinely own this worktree now, overwrite its .claude-worktree-owner marker with your session id.")}}'
# Consolidation is primarily a release step; this is the between-releases floor. The metric is
# the file's LINE COUNT — the context an agent actually burns reading the log — not entry count.
# Floor 1800 keeps the whole log inside one default 2000-line Read (headroom for the reader's
# own overhead). Nudge (never fail) past it so append-only can't grow past what agents can read.
n=$(wc -l < docs/decisions.md | tr -d ' ')
echo "docs/decisions.md is ${n} lines (consolidation floor: 1800; one Read caps at 2000)."
if [ "${n:-0}" -gt 1800 ]; then
echo "::warning::docs/decisions.md is ${n} lines (>1800) — larger than agents can comfortably read in one pass. Do a consolidation pass (prune/merge superseded entries with [decisions-edit]); don't wait for the next release. See the decisions.md header."
fi
# BLOCKING (unlike docs-reminder): the mechanizable half of the "docs-update in the
# same PR" rule for the API contract (ersatztv#303 H4/H5). If a PR touches the API
# surface (ErsatzTV/Controllers/Api/** or ErsatzTV.Core/Api/**), the generated
echo "::error::This PR changes the API surface but its generated artifacts are stale. Run './scripts/update-openapi.sh && (cd web && npm run generate:api)' and commit v1.json / v1.d.ts / endpoint-index.md in THIS PR (CLAUDE.md → Conventions; ersatztv#303 H4/H5)."
exit 1
fi
echo "Generated API artifacts are in sync."
# Formatting-as-you-touch gate (ersatztv#311): verify the .cs files THIS PR changed conform to
# .editorconfig (style + charset=utf-8, i.e. no UTF-8 BOM). Scoped to changed files so it enforces
# "normalize a legacy file when you touch it" WITHOUT a big-bang reformat of the ~2500 pre-existing
# BOM files. A PR that touches no .cs skips the expensive steps and passes trivially (always reports
# a status, so it is safe as a required check).
format:
name:Formatting (changed .cs conform to .editorconfig)
echo "Verifying ${#files[@]} changed .cs file(s) against .editorconfig..."
if ! dotnet format ErsatzTV.sln --no-restore --verify-no-changes --include "${files[@]}"; then
echo "::error::One or more .cs files this PR touches don't conform to .editorconfig (formatting or a UTF-8 BOM). Run 'dotnet format ErsatzTV.sln --include <files>' and commit the result in THIS PR — the fix-as-you-touch convention (docs/contributing.md §7; ersatztv#311). Legacy files you did NOT touch are unaffected."
exit 1
fi
echo "All changed .cs files conform to .editorconfig."
@@ -5,7 +5,7 @@ Custom IPTV channel server for Jellyfin. Forked from [ErsatzTV/ErsatzTV](https:/
## Architecture
- **Language**: C# / .NET 10
- **UI**: ChicoryTV React SPA (`web/`, Vite, served at `/app`) over the REST API — the default UI; root `/` and migrated legacy routes 302 there (`ErsatzTV/LegacyUiRedirects.cs`). The legacy Blazor Server UI (MudBlazor) still serves the remaining un-migrated admin screens — playback troubleshooting, multi/rerun collections, and playlist editing depth; Blazor home = `/system/health`, reachable via the Settings → System "Classic UI" link. Media detail pages + image folder browser landed in the SPA via #141 (PR #183); its removal is #91 phase (b), gated on #145 (playback troubleshooting) and API gaps #151/#152/#153/#155 (scheduling parity #144/#162 DONE 2026-07-07: blocks/templates/decos/deco-templates/playout editors all in the SPA;#141/#158/#161/#180 also DONE)
- **UI**: ChicoryTV React SPA (`web/`, Vite, served at `/app`) over the REST API — the ONLY UI. The legacy Blazor Server UI (MudBlazor) was removed in #91 phase (b); root `/` and every legacy route now 302 to `/app`, either via an explicit redirect in `ErsatzTV/LegacyUiRedirects.cs` or the Startup catch-all fallback (any unmatched non-`/api`/`/artwork`/`/docs`/`/openapi` path → `/app`). Historical parity work: media detail pages + image folder browser landed via #141 (PR #183); scheduling parity #144/#162, #141/#158/#161/#180, #145, #151/#152/#153/#155, and the media-source write API/SPA #202 are all DONE.
- **Pattern**: CQRS via MediatR — queries/commands in `ErsatzTV.Application/`
- **Database**: EF Core (SQLite default, MySQL optional) — context in `ErsatzTV.Infrastructure/Data/TvContext.cs`
- **Media**: FFmpeg via CliWrap, SkiaSharp for logo generation
@@ -15,7 +15,7 @@ Custom IPTV channel server for Jellyfin. Forked from [ErsatzTV/ErsatzTV](https:/
| Project | Role |
|---------|------|
| `ErsatzTV/` | ASP.NET Core host, API controllers, SPA static hosting, legacy Blazor pages, DI setup |
| `ErsatzTV/` | ASP.NET Core host, API controllers, SPA static hosting, DI setup |
- **Images** (our fork, built by `.gitea/workflows/docker-build.yml` → `192.168.1.95:3000/timothy/ersatztv`): push to `main` → `:latest` + `:<sha>` (test image); push `v*` tag → `:prod` + `:<version>` + `:<sha>`. Prod deploys via**Komodo GitOps**: the`media-servers`compose in `timothy/server-management` (`docker/bumblebee/stacks/media-servers/compose.yaml`) pins the version tag (currently `26.5.0`, deployed 2026-07-07); releasing = tag here, wait for the image build, bump that pin and push (the Komodo pre-deploy hook backs up before recreating). Test container tracks `:latest`. Pipeline details: `docs/ci-cd.md`.
- **Images** (our fork, built by `.gitea/workflows/docker-build.yml` → `192.168.1.95:3000/timothy/ersatztv`): push to `main` → `:latest` + `:<sha>` (test image); push `v*` tag → `:prod` + `:<version>` + `:<sha>`. Prod's**Komodo GitOps**`media-servers`stack follows floating`:prod`; after the immutable `:<version>` candidate passes the release scans, manually deploy the stack (Global Auto Update is the daily fallback). Both paths run the fail-closed pre-deploy backup and prod-copy migration smoke before recreation. Test tracks `:latest`. Pipeline details: `docs/ci-cd.md`.
- **Read [`docs/contributing.md`](docs/contributing.md)** before non-trivial changes — it documents the established patterns (layering, CQRS handlers, LanguageExt, Blazor/MudBlazor, EF Core + dual-provider migrations, the FFmpeg pipeline, analyzers, testing) and the **deviation policy**: match the established style; diverge only with a concrete, stated reason.
- **Read [`docs/contributing.md`](docs/contributing.md)** before non-trivial changes — it documents the established patterns (layering, CQRS handlers, LanguageExt, the ChicoryTV SPA, EF Core + dual-provider migrations, the FFmpeg pipeline, analyzers, testing) and the **deviation policy**: match the established style; diverge only with a concrete, stated reason.
- **Docs-first is a HARD RULE — read before you explore**: before ANY API / SPA / E2E / parity / scheduling work, read `docs/README.md` (index) → the convention docs (`api-conventions`, `spa-conventions`, `e2e-local`, `domain-model`, `blazor-route-parity`, `decisions`). **Do NOT reverse-engineer conventions from source (Grep/Read) before reading these** — they exist precisely so you don't. Only recon the task-specific delta the docs deliberately don't freeze (a merged endpoint's exact DTO, a Blazor page's field list). **This applies to delegated subagents too**: tell each agent which doc section to read; never let one re-derive conventions from code.
- **Docs-update is part of "done" — same PR, never a follow-up**: any PR that changes a convention, adds/migrates/redirects a route, adds/changes a `/api/*` endpoint, or reverses a decision MUST update the relevant doc in that same PR:
The `docs-reminder` CI job flags a screen/route change that skips `blazor-route-parity.md`, but it's a **non-blocking** nudge — the rule is on you, not the check.
- Follow existing MediatR CQRS pattern for new features
- Domain logic in `ErsatzTV.Core`, infrastructure in `ErsatzTV.Infrastructure`
- Keep UI thin: the SPA talks to `/api/*` only; legacy Blazor pages delegate to MediatR handlers. New screens go in the SPA (`web/`), never in Blazor
- Keep UI thin: the SPA talks to `/api/*` only; controllers delegate to MediatR handlers. All UI is in the SPA (`web/`)
- Test with **NUnit** + Shouldly + NSubstitute (the existing `*.Tests` projects); xUnit is **not** used here
- **Dependencies use Central Package Management**: versions live in the repo-root `Directory.Packages.props`; csproj reference packages by name only. Add/upgrade by editing the central `<PackageVersion>` — never put `Version=` back on a `<PackageReference>` (trips `NU1008`). See `docs/ci-cd.md` → Dependency management.
- **DB migrations target BOTH providers**: a `TvContext` model change needs a migration in `ErsatzTV.Infrastructure.Sqlite`**and**`ErsatzTV.Infrastructure.MySql` — run `scripts/add-migration.sh <Name>` (does both). CI's `migrations` job enforces model-drift + apply-to-fresh-DB per provider. See `docs/ci-cd.md` → Migration integrity.
Every task that closes a Gitea issue MUST complete ALL of these before it is considered done. Use `/done <issue>` to run through this automatically.
**Merge-consent is derived from state, not asserted (`## Done-when` convention — ersatztv#303 H6 + H10).** Any issue whose PR will merge to `main` should carry a `## Done-when` section in its **issue body** — a checklist of completion criteria (always include an "adversarial review passed" box; add per-issue criteria like tests-green, docs-updated, live-E2E). Two hooks derive merge-consent from it so a premature merge is blocked *by construction*, not by memory:
-`pretooluse-merge-consent.sh` (Claude PreToolUse on the Gitea merge tool) — **auto-grants** a merge (emits `permissionDecision: allow`, so **no** redundant mechanical prompt fires) only when the PR's CI is green **and** every `## Done-when` box on the linked issue (`fixes #N`) is ticked **and** a `Review-verdict:` comment references the PR's *current head sha* (**H10**); **denies** on an unticked box, red CI, or a stale/negative review verdict; **asks** (falls back to a human prompt) when it can't derive state (no linked issue, no `## Done-when` section, no `Review-verdict:` comment yet, no creds, Gitea down). On the auto-grant (satisfied) path the derived state **is** the consent — do not also ask conversationally to merge; a separate human confirmation is warranted only when the gate **asks** (ersatztv#314). **The H10 review-verdict convention**: after an adversarial/Codex review of a PR (or its latest fix commit), post a PR comment with a line `Review-verdict: <MERGEABLE|APPROVED|BLOCKED> @ <head-sha>` — this proves the *latest* commit was reviewed, not a stale earlier diff (ersatztv#242).
-`.husky/pre-push` → `prepush-donewhen.sh` — a fail-open backstop that blocks a direct `git push origin main` whose commits `fix #N` an issue with unticked boxes.
Both need Gitea read creds in the env to enforce (**`ETV_GITEA_BASICAUTH=user:pass`** or `ETV_GITEA_TOKEN`; `ETV_GITEA_URL` overrides the base). Without them the merge hook asks and the push backstop is a no-op — the gate degrades to today's manual confirmation, never a silent pass. Docs-only PRs/pushes are exempt.
1.**Root cause** (bug fixes / incidents only): Document WHY the problem existed, not just what was changed. If root cause is unknown, say so explicitly and open a follow-up investigation issue. Fixing symptoms without understanding causes creates recurring problems.
2.**Comment on issues** as you work — what you found, what approach you're taking, any deviations from the suggested fix.
3.**Push changes**: `git push` all commits before closing. Use `fixes #N` in commit messages to auto-close where appropriate.
returnBaseError.New($"Block name \"{request.Name}\" is invalid");
}
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.