The retention purge job gated its whole body on activeRetention while
every collection declares only postDeletion, and no app ever called
registerRetentionPurgeJobs — retention was dead end to end (audit
findings A2 + A3). The DSR soft delete now stamps a deletedAt tombstone
on postDeletion collections (kept distinct from processingRestrictedAt
so an Art. 18 restriction never feeds the purge), the job grace-purges
tombstoned rows past postDeletion.duration with the declared action,
core-cms injects the tombstone field + Payload task definitions, and
bindAllProduction enqueues the first purge cycle at boot.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Adds a shared requireAuthenticated tRPC middleware (reads the server-
resolved ctx.user from createTrpcContext) and applies it to every
mutating feature procedure — blog.createArticle and media.deleteMedia
were anonymous-callable (audit finding B7). Read-only queries stay
public; features compose <x>ProtectedProcedure from their error-mapped
base procedure.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The tRPC createContext was () => ({}) — the mounted dsr/consent routers
401'd every call and the dsr singleton stub threw (audit finding A11).
createTrpcContext now accepts an app resolveUser hook; web-next resolves
the session cookie through the auth feature's validateSession (denylist
included) plus a role snapshot, and threads bindProductionDsr/Consent
(or dev-seed) bindings into every request. The dsr router resolves its
binding from ctx.dsrBinding per request instead of a throwing proxy.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The router previously created its own initTRPC without superjson while
the app router uses the shared transformer-enabled instance — a wire
transformer mismatch that corrupts inputs (audit finding A10). Procedures
now build on @repo/core-shared/trpc/init's t; a real client+fetch-adapter
round-trip test pins Date revival through appRouter-style mounting.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Non-admin callers may export/rectify/restrict/delete ONLY their own
subjectId; a mismatch is rejected with FORBIDDEN instead of being
honored verbatim (IDOR, audit finding A1). Cross-subject operations
require the admin role; cascade-hard stays admin-only on top.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@trpc/client resolved to 11.16.0 in some packages while @trpc/server
resolved to 11.17.0 in others, tripping the exact-version peer check.
Update every @trpc/* range in-range so the whole workspace resolves to
one version (11.18.0); install is peer-warning-free again.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
core-trpc imported AppRouter (type-only) from core-api, which depends
on the features, which depend on core-trpc — a cycle that killed every
turbo graph walk (lint/build/CI) and an illegal core -> core-composition
boundary edge. The tRPC context is now router-agnostic (AnyTRPCRouter);
useTRPC takes the router type as a generic and each feature exports a
type-only app-slice (e.g. BlogAppSlice) mirroring its mount key, so UI
hooks keep full procedure typing without touching the composition layer.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
IHeaderRepository.getHeader() now returns Header | null, making the
use case's HeaderNotFoundError branch honestly reachable instead of
dead code behind a lying type (B12). The Payload repo also skips CMS
rows with empty label/href rather than emitting items that violate
headerItemSchema.min(1) and 500 at output validation.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
getPageBySlug returned undefined while the feature already mapped
PageNotFoundError to NOT_FOUND and blog throws for missing slugs (B11).
Unify on the throwing contract: use case throws, controller narrows to
a non-optional presenter, the server component catches the domain error
to render its not-found state, and the router now surfaces NOT_FOUND.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Same B9 defect as blog: Payload paginates by page, so a non-aligned
offset returned the wrong window. Fetch the straddled pages and slice
the intra-page remainder. Identical window tests now run against the
mock and the fake-payload stub.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Payload paginates by page, so floor(offset/limit)+1 alone returned the
wrong window whenever offset % limit != 0 (B9). Fetch the straddled
pages and slice the intra-page remainder. The test stub now honours
page like real Payload (it silently ignored it before), and the
contract pins aligned + non-aligned windows on mock and stub alike.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
clientIp was part of the public signInInputSchema, so any client could
spoof its own rate-limit bucket or dodge IP throttling entirely (audit
finding B2). The schema no longer carries it (strict parsing rejects it
with BAD_REQUEST); instead the web-next tRPC fetch adapter derives it in
createTrpcContext from x-forwarded-for (first hop) / x-real-ip — trust
caveat documented — and the router threads ctx.clientIp to the
controller as a second, server-only argument typed outside the input
schema (SignInRequestContext).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The session methods had zero coverage because they call getPayload()
(audit finding B8). Stub the payload module (secret + findByID only) and
exercise the real crypto paths: round-trip, tampered signature, swapped
payload, expired exp, malformed segments, wrong secret, jti-less token
(fail closed), post-revocation rejection, and the per-instance denylist
limit. No running Payload needed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
invalidateSession previously ignored its argument, so a stateless JWT
stayed valid until exp after logout (audit finding B5). createSession
now embeds the minted session.id in the token as jti; invalidateSession
records that jti in an in-memory denylist with expiry-based pruning, and
validateSession rejects denylisted (and jti-less, failing closed)
tokens. The denylist is per-process — the single-process limitation and
the shared-store upgrade path are documented in session-denylist.ts.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The package's vitest base configs were outside every gate — no tsconfig
and no typecheck script meant tsc never saw them. Add a tsconfig with
allowImportingTsExtensions (vitest.base.jsdom imports ./vitest.base.node.ts
by extension) and an explicit include of the base configs + vitest
config, plus the typescript/@types/node devDeps to run it. (S6)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Next.js only injects the nonce into its own scripts when it can read it
from the request's Content-Security-Policy header. Setting the CSP only
on the response left hydration scripts un-nonced in production (A9).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
getUserByUsername emitted the FULL username as the emailDomain span
attribute whenever the username contained no "@" (audit finding B6),
violating the PII-free telemetry rule (ADR-017 §7). Emit only a boolean
hasAtSign in both the production repository and its mock; regression
tests pin that no username-derived string reaches span attributes.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Handlers and the published record now receive the zod-PARSED payload
(defaults/coercions applied) instead of the raw input, and fan-out uses
Promise.allSettled with errors swallowed by default plus an opt-in
failFast — matching the InMemoryEventBus the events core-package
generator scaffolds. The previous sequential fail-fast divergence was
undocumented, so it is aligned rather than kept. Regression tests pin
parsed-payload delivery, non-short-circuiting fan-out, and failFast
rethrow. (S4; the generator bus templates already carry the fix.)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The production UsersRepository reads and writes username + passwordHash
via the Payload local API, but the users collection never declared them,
so production sign-up/sign-in was broken (audit finding B1). passwordHash
uses access.read: () => false so credential material never serializes
through any Payload API surface; the repository still reads it with
overrideAccess: true. A contract-shaped test pins every repo-used field
(USERS_REPOSITORY_FIELDS) against the collection config so drift fails
at test time without a database.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
setup/node.ts installs the no-instrumentation guard (vi.mock against
real Sentry/OTel SDK init) via its import — it was never a no-op. Fix
the same lie in core-typescript's jsdom base-config test comments and
test title. (S2)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
pnpm audit --prod --audit-level=critical (the new CI gate) failed on
vitest <3.2.6 (critical, arbitrary file read/execute via the UI
server). Bump vitest + @vitest/coverage-v8 in-range across the
workspace and add @vitest/coverage-v8 to every package that runs
vitest but lacked it (core-audit, core-cms, core-eslint, core-testing,
core-trpc, apps/cms, web-next, web-tanstack, turbo/generators) so
'pnpm test -- --coverage' works in every package. No compound test
scripts exist, so the vitest-last pass-through concern does not apply.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
*.stories.tsx / *.stories.ts were excluded from tsconfig, so story
files never typechecked and could drift against component props
silently. Include them (they already typecheck cleanly); build is
tsc --noEmit so nothing new is emitted. The Select scrollIntoView port
from the downstream fork does not apply — this tree has no Select atom.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The previous init-server-node tests only checked that an SDK handle
came back — the ADR-017 §7 invariant (PiiScrubSpanProcessor /
PiiScrubLogRecordProcessor registered BEFORE any exporter-facing
processor) was untested. Capture the NodeSDK constructor config via a
local vi.mock override and assert processor ordering for both the
empty-DSN and DSN-set paths.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
httpBatchLink joins same-tick calls into one request whose path is a
comma-separated list of procedure paths; the stub matched the joined
string against a single mock key, so any batched pair failed with 'No
mock for a.one,a.two'. It now answers one element per procedure, in
order. Error bodies are also superjson-serialized — raw error JSON made
the client throw 'Unable to transform response' instead of surfacing
the intended error. Regression tests added.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
next.js and react-internal.js are plain re-exports of base, so no
rules-of-hooks checking was active anywhere despite three React apps
and core-ui. Wire eslint-plugin-react-hooks in base.js scoped to
**/*.tsx (non-React packages untouched). Fixes the one violation it
surfaced: web-tanstack's root route called Route.useLoaderData inside
an anonymous component callback — extracted to a named RootComponent.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
ProductionUseCase<I, O, M> only demanded Instrumented + Captured (+
Audited for mutating-with-audits). The boot assertion additionally
requires __analyzed for non-empty analyticsEvents and __rateLimited for
non-empty rateLimit, so the type-level gate under-promised what boot
enforces. The slot now derives both from the manifest entry; the
feature-scoped requiresConsent brand stays boot-only (documented).
Also make IAnalytics extend AnalyticsProtocol from
core-shared/di/bind-protocols so narrowing the ctx protocol fails
typecheck in core-analytics instead of drifting silently.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- turbo/generators/config.ts used require() inside the reader generator
action (no-require-imports); use the top-level node:fs imports
- auth authentication.service.ts used a literal self-assertion
("users" as "users"); prefer-as-const
- apps/cms lacked web-next's next-env.d.ts triple-slash-reference
override, and the committed next-env.d.ts now references
.next/types/routes.d.ts
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The base config documented (and carried allowlist off-blocks for) the
OTel-SDK import restriction but the no-restricted-imports pattern group
only covered @sentry/*. Add @opentelemetry/sdk-*, exporter-*,
instrumentation-*, resources and semantic-conventions to the restricted
group, and extend the off-block for core-audit's trace-id enrichment
test, which legitimately builds an in-memory sdk-trace-base tracer.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
usecase-must-be-wired returned {} when the manifest parsed to null,
silently disabling the error-level gate when the manifest existed but
could not be read. It now reports unparseableManifest on Program in
that case; a genuinely missing manifest stays a no-op (that is
feature-must-have-manifest's job).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A manifest written as `{...} satisfies FeatureManifest` (or the combined
`as const satisfies` idiom) parsed to null, silently no-oping the
error-level conformance rules. unwrapAsConst now strips TSAsExpression
and TSSatisfiesExpression in a loop. The file also carried a verbatim
second copy of its own parser for parseManifestFully; both public entry
points now share one implementation. The template's field set (audits,
rateLimit, requiresConsent) is preserved.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Replace NotImplementedError stubs in AuthenticationService with working
implementations: createSession signs a HS256 JWT using Payload's instance
secret, validateSession verifies and decodes the token then looks up the
user, invalidateSession returns a blank cookie with maxAge 0. No external
JWT dependency — uses Node crypto HMAC directly.
Also clarify withAudit/withAnalytics comments: the wrappers intentionally
delegate recording to the use case body (only it knows which fields to
extract), so the TODO was misleading.
- Add ReadOnly<F> phantom brand to core-shared/conformance (compile-time
enforcement that readers only wrap non-mutating use cases)
- Add isReadOnly runtime predicate for boot-time assertReaderPurity
- Scaffold pnpm turbo gen reader: creates integrations/readers/ with
interface, implementation, test, barrel, and adds ./reader export
subpath to package.json
Add reads field to UseCaseManifest, update CLAUDE.md with Q0-Q3 rules,
add ./reader subpath to AGENTS.md exports table, and cascade reader
conventions through conformance quickref, adding-a-feature guide, and
scaffolding guide. Moves gen reader from deferred to planned.
Server components (.server.tsx) resolve controllers from DI, prefetch
data, and wrap client components in HydrationBoundary. Client components
(.client.tsx) use hooks for hydration + background refetch. Barrel
exports server components under clean names — consumers never see the
server/client split.
- Split globals.css into globals.css + theme.css so Storybook can
import design tokens without the @import "tailwindcss" directive
that breaks Vite's preview module loading
- Prepend @tailwindcss/vite plugin for correct processing order
- Add @source to scan core-ui components for utility class generation
Webpack (Next.js transpilePackages) resolves from .ts source directly
and cannot find .js files. The bind-production binders already used
extensionless imports; align bind-dev-seed to match. Also fixes the
turbo gen feature template so new features are consistent.
turbo.json's boundary config already allows `feature -> feature`, and
the cross-feature event system depends on it — a consumer must import
the publisher's event contract from `@repo/<publisher>`. But the ESLint
boundaries config, ADR-010, and AGENTS.md still declared
`feature -> [core, tooling]`, contradicting turbo.json and the shipped
code (marketing-pages imports @repo/auth).
Align all three to turbo.json: a feature may import another feature's
published public exports. Internals stay sealed by the `exports` map,
and cross-feature behaviour still flows through IEventBus.
Two CLAUDE.md conventions had no mechanical gate, so both drifted:
entity models shipped without sibling tests, and feature test files
imported src modules via `../` instead of the `@/` alias.
- `entity-must-have-test` — every entities/models/<x>.ts needs a sibling
<x>.test.ts (errors and barrels excluded).
- `no-relative-parent-import-in-tests` — feature test files must import
src via `@/`, not `../`. Scoped to feature packages; core packages are
governed by their own generator templates.
Both register at warn level, bringing the conformance rule count to 15.
The rule only matched bus.publish("string-literal", ...), but the
canonical pattern that `gen event` prescribes is
bus.publish(eventDescriptor, payload) — an imported identifier, never a
literal. The rule therefore never fired on real code, which is how the
auth signUp publish drifted from its manifest undetected.
Add `_event-ast.js`: resolves a `bus.publish(<identifier>)` argument by
following the import to the event-contract file and extracting the name
from either `defineEvent("...", schema)` or an inline `{ name }` object.
Unresolvable arguments are skipped, so the rule never false-positives.
bind-production.types.test.ts imported feature.manifest and the sign-in
use case with `../` parent paths. Test files import src modules via the
`@/` alias per the repo convention — this clears the way for the new
no-relative-parent-import-in-tests rule.
container.test.ts imported cross-directory modules with `../...` paths.
Test files use the `@/` alias for src imports per the repo convention;
this was the outlier vs auth/media/navigation.
cookie.ts was the only entity model in the feature packages without a
sibling test. Covers minimal/full attributes, invalid sameSite, and a
missing required field.
The signUp use case calls `bus.publish(userSignedUpEvent, ...)` but its
manifest entry declared `publishes: []`. The conformance event graph was
blind to the entire auth -> marketing-pages welcome-email flow as a
result. Declaring "auth.user.signed-up" makes `pnpm conformance` surface
the event and its publisher.
Register core-shared/security/tanstack server middleware in app.config.ts
as a Nitro/H3 hook that emits the six security headers and forwards the
per-request nonce. Update instrumentation-client to read the nonce from
<meta name="csp-nonce"> and pass it to initSentryClientReact.
Add nonce support to initSentryClientReact (feedbackIntegration receives
styleNonce/scriptNonce), mirroring the initSentryClient pattern already
in place for web-next.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add apps/web-next/middleware.ts calling withSecurityHeaders() from
core-shared/security/next; exports matcher config excluding static assets
- Update layout.tsx to call getNonce() and render <meta name="csp-nonce">
so client-side JS can read the per-request nonce
- Update instrumentation-client.ts to read nonce from csp-nonce meta tag
and pass it to initSentryClient for feedbackIntegration CSP compliance
- Add nonce option to initSentryClient (InitClientOpts.nonce) and thread
styleNonce + scriptNonce into feedbackIntegration when provided
- Add middleware test asserting all six headers, prod/dev CSP shape, and
x-nonce presence; add feedbackIntegration nonce tests to core-shared
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
