문제
guardian은 fd 0을 fs.createReadStream으로 읽는다. 이 스트림은 libuv
스레드 풀의 스레드 하나를 read(2)에 세워 둔다. process.exit()는 종료
전에 스레드 풀을 join하므로, 그 read가 돌아올 때까지(부모가 stdin을
닫을 때까지) 끝나지 않는다. 그래서 가드를 거부하거나 link가 실패하면
failClosed가 진단을 바로 쓰고도 프로세스는 남는다. 클라이언트는 stdin을
연 채 READY를 기다리므로 거부가 즉시 오지 않고 고정 5초 뒤 "READY timed
out"으로 끝난다. 같은 종류 guardian 두 개 테스트의 패자가 이 경로다.
탐침: 같은 종료 코드를 fs 스트림은 stdin을 닫은 2,008ms에, net 소켓은
228ms에 냈다(200ms 뒤 exit 호출).
고른 방법
- liveness 입력을 process.stdin으로 바꾼다. pipe면 Node가 이벤트 루프로
읽는 net.Socket을 만든다. 스레드 풀 read가 없어 exit가 바로 끝난다.
data/end/error 처리, removeAllListeners·destroy, closeSync(0)는 그대로다
(libuv는 fd 0~2를 destroy에서 닫지 않는다).
- 동시 guardian 테스트에 붙였던 PROCESS_HEAVY_TIMEOUT_MS와 그 주석을
뺀다. 패자가 link 실패 즉시 끝나 테스트가 5.14초에서 0.16~0.54초가 됐다.
버린 대안
- provider-scope-wrapper.ts처럼 new Socket({ fd: 0 }): fd 0이 pipe가
아니면 생성자가 throw한다. guardian은 그 시점에 이미 private alias를
묶었고 failClosed는 liveness가 있어야 돈다. process.stdin은 pipe에서는
같은 Socket이고, 그 밖의 fd에서는 throw 대신 end/error로 기존 실패
경로(125)를 탄다.
- 실패 때 SIGKILL로 자신을 죽이기: 종료 코드 124~126 계약이 사라진다.
위험·한계
- pipe인 fd 0에 O_NONBLOCK이 걸린다. 읽기 끝은 guardian만 가진다
(클라이언트는 stdio "pipe"로 띄운다).
- lease 마감 경로는 원래 SIGKILL을 쓰므로 바뀌지 않는다.
검증
- RED(수정 전): 새 테스트 "exits on a rejected guard while the parent
still holds its input open"이 "rejected-guard guardian exit timed out"
(5,122ms)으로 실패. 다른 nonce의 가드를 보내 시계와 무관하게 거부시킨다.
- GREEN: tests/integration/provider-guardian-transaction.test.ts 22/22
통과, 3회 반복 통과(새 테스트 65~333ms). tests/unit/task3-selective-
integration 7/7. tsc(tsconfig.node, tsconfig.test) 통과, 바꾼 파일
eslint 통과.
- 미실행: tests/system/ci-runner/ci-artifact-contract.test.ts의 provider
체인 테스트. fixture가 corepack pnpm으로 빌드해서 이 worktree에서는
돌리지 않았다.
Refs: commit c862004 위험·한계의 "따로 남는 결함"
문제
guardian 테스트는 자식을 띄우자마자 절대 마감(Date.now() + 350)이 든
가드를 보낸다. 그래서 lease가 프로세스 기동 시간까지 낸다. 1코어에서
기동이 350ms를 넘으면 guardian이 가드를 마감 지남으로 거부하고, 테스트는
5초 뒤 "deadline guardian READY timed out"으로 실패한다(수정 전 1코어
재현 9회 중 4회). 같은 구조가 2~3초 lease를 쓰는 테스트 5곳에 있다. near-timeout
테스트는 READY 뒤 1초를 고정으로 자서, 기동이 길면 publish·commit에 남는
시간이 줄어든다. 자식 기동을 기다리는 poll 세 개(waitForFile,
waitForNewDirectChild, waitForStoppedProcess)는 1초 마감을 쓴다. 파일
머리의 watchdog 설명("1초 요구를 넣지 않는다")과 어긋난다.
고른 방법
- spawnGuardian이 실제 guardian을 import하는 래퍼로 띄운다. import가
끝나면(bootstrap과 stdin reader 등록까지 끝난 시점) 래퍼가 여분 pipe
(fd 7)에 1바이트를 쓰고 닫는다. 새 sendGuard는 이 신호를 받은 뒤에
마감을 계산해 가드를 보낸다. 가드를 보내는 6곳이 모두 sendGuard를 쓴다.
guardian 코드와 프로토콜, 350ms·2s·3s lease 값은 그대로다.
- near-timeout 테스트는 startProviderGuardian 호출 직전 시각에서 1초 뒤에
깬다. 클라이언트가 spawn 뒤에 마감을 찍으므로 마감은 그 시각 + 2초
이후다. 기동이 길어도 publish·commit에 1초가 남는다.
- 1초 poll 세 개는 같은 파일의 watchdog(5초)을 쓴다. 자식 기동에는 기다릴
이벤트가 없어 poll이 남는다.
- 같은 종류 guardian 두 개 테스트는 패자가 클라이언트의 고정 5초 READY
제한(RESPONSE_TIMEOUT_MS)으로만 거부된다. 테스트가 줄일 수 없는 고정
대기라 파일의 PROCESS_HEAVY_TIMEOUT_MS를 붙이고 이유를 주석에 적었다.
버린 대안
- lease를 늘리기: 기동 시간에 상한이 없어 같은 문제가 더 드물게 남는다.
350ms 테스트는 만료를 기다리는 시간도 길어진다.
- marker 파일 + poll로 준비 신호: 파일 시스템 poll에 다시 마감이 필요하다.
pipe는 이벤트로 끝나고 provider-evidence 목록 단언에 끼어들지 않는다.
- guardian 프로세스에 가짜 시계 주입: 자식 프로세스라 테스트용 훅을 운영
스크립트에 넣어야 한다.
위험·한계
- spawnGuardian 경로의 guardian은 main 모듈이 아니라 래퍼가 import한
모듈로 돈다. guardian은 argv·import.meta를 읽지 않는다. 래퍼는 신호를
보낸 뒤 fd 7을 닫으므로 가드를 받는 시점의 상속 fd(0~6)는 운영과 같다.
클라이언트 경로 테스트는 그대로 실제 스크립트를 직접 띄운다.
- 가드를 보내기 전 기동 대기는 watchdog 5초가 상한이다. 남은 실제 시계
구간은 가드 처리(decode, link, fsync, READY)뿐이다.
- 따로 남는 결함(이 커밋 범위 밖): guardian은 가드를 거부하거나 link가
실패하면 failClosed가 진단을 쓰지만, stdin 읽기가 걸려 있는 동안
종료하지 않는다. 탐침: 만료된 가드를 받고 1,008ms에 진단을 쓴 뒤 stdin을
닫은 4,014ms에야 126으로 끝났다. 그래서 클라이언트에서는 거부가 즉시
오지 않고 5초 뒤 "READY timed out"이 된다(leaseMs 1과 동시 guardian 탐침
5,018~5,053ms). 위 PROCESS_HEAVY_TIMEOUT_MS는 이 결함이 고쳐지면 뺄 수
있다.
검증
- RED(수정 전, taskset -c 0, --maxWorkers=4): component·features·
integration·platform/cardinality 3회 중 2회, 같은 묶음 + 바쁜 이웃 1개
3회 중 2회 "expires an uncommitted lease … deadline guardian READY timed
out"(5,035~5,052ms).
- GREEN(세 커밋을 합친 상태, taskset -c 0 + 같은 코어의 바쁜 이웃,
--maxWorkers=4): 대상 6개 파일 6회, 위 넓은 묶음 5회(이웃 1개) 모두
통과. 이웃 2개로 3회 돌렸을 때 이 파일은 모두 통과했다(실패 1건은
http-execution-v3-live-authority, 0929/10-03).
- 기본 코어: tests/unit·component·integration 212파일 2,734개 통과
(pnpm을 띄우는 security-followup-archive, fixture-node-modules,
lab-performance 제외). tsc -p tsconfig.test.json, 바꾼 파일 eslint
통과.
Refs: docs/reviews/2026-11-01-maturity-review/10-tests-fast-layers.md
문제
참조 기능의 생성 분류기가 모든 409를 NOT_APPLIED로 판정한다
(reference-feature-contract-contribution.ts:42-44). 결과 불명 쓰기를 같은
멱등 키로 다시 보냈을 때 첫 요청이 아직 처리 중이면 짝 백엔드 템플릿은
409 IDEMPOTENT_IN_FLIGHT를 준다. 그러면 코디네이터가 intent를 정리하고
conflict를 세우고(application-query.ts:584-589), 폼은 결과 불명 기록을
지운다(use-app-form.ts:227-230). 화면이 "충돌"로 바뀌고 저장 버튼이 다시
열려 다음 저장이 새 키로 나간다. 첫 요청이 끝나면 같은 리소스가 두 번
생긴다(키 [K, K, K']).
고른 방법
분류기가 409를 problem code로 나눈다. fixture가 선언한 충돌 code
(CREATE_NOT_APPLIED_CONFLICT_CODES, 지금은 MSW fixture가 내는 "CONFLICT"
하나)만 NOT_APPLIED다. IDEMPOTENT_IN_FLIGHT, code 없는 409, 모르는 code의
409는 MAYBE_APPLIED다. 결과 불명이 유지되므로 코디네이터가 intent를 들고
있고 다음 전송이 같은 키를 쓴다. 400·422 판정은 바꾸지 않았다.
시나리오 카탈로그에 in-flight-409를 더해 "처리 중 409 -> 결과 불명"과
"충돌 409 -> 적용 안 됨"(기존 conflict-409)을 둘 다 고정했다.
버린 대안
- IDEMPOTENT_IN_FLIGHT 하나만 MAYBE_APPLIED로 빼기: 모르는 409가 계속
NOT_APPLIED다. 백엔드가 처리 중 code를 바꾸거나 다른 백엔드를 붙이면
같은 결함이 조용히 돌아온다. "모르는 409는 안전한 쪽" 결정과 어긋난다.
- 플랫폼 방어(같은 intent의 재전송이 NOT_APPLIED를 받아도
APPLIED_CONFIRMED 전까지 키 결합을 유지, 리뷰 03 5절 A의 (b)): 분류기가
틀려도 막지만 application-query의 admission 수명 규칙을 바꾸는 구조
변경이다. 이번 범위를 넘어 뺐다.
- 계약 문서의 409 IDEMPOTENCY_KEY_REUSED(같은 키·다른 지문)를 NOT_APPLIED
목록에 넣기: 프론트는 canonical 입력이 같을 때만 같은 키를 쓴다. 이
응답은 서버와 클라이언트의 지문 기준이 어긋났다는 뜻이고 같은 키의 앞선
요청은 적용됐을 수 있다. NOT_APPLIED로 풀면 다음 저장이 새 키로 중복을
만든다. 그래서 모르는 409와 같이 MAYBE_APPLIED로 둔다.
위험·한계
- 백엔드 계약(이 repo에서 끝낼 수 없음): backend-api-and-server-state-
contract.md §6은 같은 키 동시 재전송을 "bounded IN_PROGRESS 결과"라고만
적고 status·code를 정하지 않는다(:299-300). 409 + IDEMPOTENT_IN_FLIGHT로
고정하고, 진짜 충돌을 뜻하는 409 code 목록을 계약에 적어야 한다. 짝
백엔드 템플릿(01a8a2a, 읽기만 함)은 OperationalError.IDEMPOTENT_IN_FLIGHT
(CONFLICT, 409, retryable=false)를 쓴다. 승인자: 템플릿 소유자, 백엔드
템플릿 소유자.
- problem 형태: 짝 백엔드 web 어댑터는 실패를 problem+json이 아니라
Envelope {success:false, error:{code,...}}로 보낸다(ErrorResponseFactory).
참조 fixture는 최상위 code가 있는 problem 형태다. 그 Envelope가 오면
referenceProblemSchema 검증이 실패해 MAYBE_APPLIED가 되므로 안전 쪽이지만
code 분류는 닿지 않는다. 제품 계약 패키지가 code 위치를 정해야 한다.
- 화면 종류는 여전히 mapProblem이 409 전부를 CONFLICT로 만든다
(reference-http-gateway.ts:38). 폼이 effect를 먼저 보므로 결과 불명이
이기지만, 효과와 화면 종류가 두 함수에 나뉜 문제(03-04)는 그대로다.
"처리 중" 전용 문구는 없고 일반 결과 불명 문구를 보인다.
- 다운스트림 tech-log 분류기는 4xx 전체를 NOT_APPLIED로 쓴다(지난 라운드
확인). 동기화 때 같은 수정이 필요하다.
- 카탈로그 실행 수가 54 -> 57이 된다. 문서 세 곳의 "19개"와
run-http-scenario-evidence.ts 출력 문구를 맞췄다.
- 초기 JS gzip +37 B(203,507 -> 203,544, 예산 204,800). 같은 build-frontend로
두 번 빌드해 gzipSync로 쟀다. 예산은 바꾸지 않았다.
검증
- RED(수정 전): reference-production-vertical "keeps the key when the
resend meets the first attempt still in flight" 실패, 키 [K, K, K'].
분류 단위 테스트 3건 "expected 'NOT_APPLIED' to be 'MAYBE_APPLIED'"
(IDEMPOTENT_IN_FLIGHT, 모르는 code, code 없음).
카탈로그 CREATE_REFERENCE_RESOURCE::in-flight-409 불일치.
- GREEN: 같은 세 파일 23/23 통과.
- 회귀: run-vitest tests/features tests/component tests/contract
tests/integration tests/unit(pnpm을 띄우는 security-followup-archive
제외) 233 files / 2817 tests 통과. tests/runtime-schema 56 통과.
tsc app/test/node rc=0. 바꾼 파일 eslint rc=0. check-adapter-inventory
PASS. verify-documentation-readiness PASS_SCOPED.
- 미실행: check:architecture(pnpm 필요). depcruise 직접 실행은 TS 7을
읽지 못해 0 modules라 사실상 미검증. src 변경은 import를 더하지 않는다.
check:ci 미실행.
Refs: docs/reviews/2026-11-01-maturity-review/03-adapters-http-state.md
pnpm verifies the install before a script, and a fixture running it with
CI=true against an install made without CI saw other settings and installed
through the linked .pnpm and .bin, rewriting the shared lock.yaml and shims.
Every fixture pnpm run now turns that check off; a test pins it for all.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A bare vitest run collected 26 Playwright specs and the host-bound system
suite. The documented one-file command goes through run-vitest.ts, which
renders the gitignored product index a fresh clone lacks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With no credential V3 sends nothing and returns UNAUTHENTICATED without a
status, yet it was counted as an API failure, so signed-out states raised
the failure rate. Diagnostics still record it once; telemetry does not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
V3 read the response x-correlation-id and dropped it, and untraced requests
sent no id, so a support reference could not be matched to a server log.
Each logical execution now sends one x-correlation-id; failures carry the
echoed value as requestId and diagnostics record it as correlation_id.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An external owner without readSubject made every authenticated
notification (every token refresh) abort requests and clear the
server-state cache. readSubject is now a required owner method and
getSubject a required SessionGateway method.
- ExternalSessionOwner.readSubject(): required; the host owner check
(EXTERNAL_OWNER_METHODS) rejects an owner without it, so the session
fails closed as integration-failed like any other missing method.
- SessionGateway.getSubject(): required. Anonymous and unavailable
adapters return null; demo already returned its fixed user.
- Scope runtime: same (state, subject) still skips the fence; a state
or subject change still fences. An authenticated session with a
null/empty subject still fences (fail closed) and now records
cache.operation.failed {operation: SCOPE_FENCE, reason:
SUBJECT_MISSING} once per fence. Bootstrap passes diagnostics in.
- Docs: starter-experience external owner section and port comments
describe what readSubject returns, the fail-closed diagnostic and a
migration note.
RED before the change (vitest tests/unit/server-state-scope-runtime.test.ts
tests/unit/runtime-adapters.test.ts): 4 failed | 31 passed
- fails closed when the owner lacks readSubject:
expected 'authenticated' to be 'integration-failed'
- fences and reports SUBJECT_MISSING ... null / empty subject, and the
external owner whose readSubject returns null:
expected [] to deeply equal [[{ level: 'warn', ... }]]
- type level: an owner literal without readSubject compiled (tsc exit
0). The new @ts-expect-error pin fails check:types:test with TS2578
if readSubject becomes optional again.
Pins that stayed green: same-subject token refresh keeps the cache,
subject change fences, sign-out fences, owner missing readState is
rejected.
GREEN: check:architecture, check:types:app, check:types:test, lint,
test:unit (2084), test:component (182), test:reference-feature (59),
test:integration (151), verify:documentation, check:diagnostics,
check:ci all pass.
BREAKING: external auth owners must implement readSubject(); code that
implements AuthSessionPort/SessionGateway directly must implement
getSubject().
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- client: the final V2 classification step maps a 408/425 response with a
valid error envelope to SERVER_FAILURE (via gatewayFailureOr), keeping
the backend code and request/trace ids. contracts/errors.ts is unchanged.
- client: Retry-After is also parsed (kernel parser) on 503, so the retry
waits at least that long and the 30s ceiling applies. 502/504 keep the
local backoff.
RED (before the fix, 14 failing in tests/integration/http-client.test.ts):
- safe GET on a 408/425 error envelope: kind UNKNOWN_CLIENT_FAILURE,
fetched once (expected SERVER_FAILURE, 3 fetches, code UPSTREAM_BUSY)
- none/keyed command on a 408/425 error envelope: kind
UNKNOWN_CLIENT_FAILURE; keyed fetched once instead of 3 times
- 408/425 error envelope with Retry-After: 2: not retried
- 503 (gateway body and envelope) with Retry-After: 2: slept 125ms
instead of 2000ms
- 408/425 envelope and 503 with Retry-After: 31: retryAfterMs missing
Pins that passed before and after: a non-idempotent command is sent once
on 408/425/503 gateway bodies and on a 503 envelope; 502/504 ignore
Retry-After (125ms).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- retry-policy: a SERVER_FAILURE with status 408/425 follows the
502/503/504 rule (safe or keyed only; never idempotency "none" or
retry "never"). Retry-After is honored and its 30s ceiling applies to
any failure that carries it, not only RATE_LIMITED.
- client: Retry-After is parsed by platform/retry-after.ts for
408/425/429, on every failure path (gateway bodies included).
The local parseRetryAfterHeader is removed.
RED (before the fix, 15 failing):
- 408/425 gateway body on a safe GET: expected 3 fetches, got 1
- keyed command on 408/425: expected 3 fetches, got 1
- 408/425 with Retry-After: 2: expected ok after a 2000ms wait, got the
first failure
- 408/425 with Retry-After: 31: retryAfterMs missing on the failure
- 429 Retry-After "1.5": slept 1500ms instead of the 125ms backoff;
"Thursday, 01-Jan-70 00:00:05 GMT": slept 5000ms; "1e3", "0x10",
"-1", "2099-01-01T00:00:00Z": accepted by the old parser, so the
request was not retried with the 125ms backoff
- unit shouldRetry(safe, SERVER_FAILURE 408): false
Non-idempotent command on 408/425 stays 1 fetch (pinned, passed before
and after).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
V2 checked the content type and decoded the body before looking at the
status. An nginx 502 with an HTML page became CONTENT_TYPE_MISMATCH and
an empty or broken 503/504 JSON body became MALFORMED_JSON: both
non-retryable and shown as a contract breach. For a 5xx/408/425 status
these branches, and the invalid-envelope branch, now return
SERVER_FAILURE with the httpStatus, the same R-3.3 rule V3 uses.
Retry follows V2's existing SERVER_FAILURE rule in retry-policy.ts, so
502/503/504 retry and 408/425/500 do not; SERVER_FAILURE was already in
retryKinds, so retry-policy.ts is unchanged.
The existing "non-JSON response" test used a 502 and pinned the old
CONTENT_TYPE_MISMATCH classification; it now uses a 200, which still
covers the non-JSON rejection and body redaction.
Verified: four new http-client cases (502 HTML, 503 empty JSON, 504
broken JSON, 408 text) failed RED with CONTENT_TYPE_MISMATCH /
MALFORMED_JSON and now pass with the expected fetch counts. All lane
gates pass except the known removal-fixture test; test:reference-feature
timed out once under load and passed on rerun.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The V2 client only attempted 401 session recovery while the physical
attempt count was under maxRetryAttempts + 1. With MAX_RETRY_ATTEMPTS=0,
or when a 503 retry had already used the budget, a recoverable 401 went
straight to onUnauthenticated() and logged the user out. Recovery is
already bounded to one replay by recoveryUsed and by the deadline, so
the budget check is removed.
The comments in runtime-adapters.ts and adapters/http/index.ts said V2
works as a rollback path for V3 once its four slots are injected. That
is wrong: V2 validates the {success,data,meta} envelope and V3 the bare
DTO, so the same backend fails every V2 response with
ENVELOPE_MISMATCH. The comments and the client.ts ledger row in
docs/reviews/adapters/01-network-and-state.md now say a rollback also
needs a V2-envelope server. No docs/architecture page made the claim.
V2 code is kept.
Verified: two new auth-recovery cases (retries disabled; 503 then 401)
failed RED with ok:false and now recover. All lane gates pass except
the known removal-fixture test; verify:documentation passes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The shared credential admission (V2 and V3) only rejected CR/LF. A token
with NUL, a non-Latin-1 character or another control character passed,
then Fetch rejected it after V3 had marked the attempt dispatched: a query
burned its retry budget and a command ended MAYBE_APPLIED with
NETWORK_FAILURE. The external session adapter also merged header names
that differ only in case, silently keeping the last value.
- admitCredentialHeaders: the value must match [\x20-\x7E\xA0-\xFF]+ (a
ByteString without C0/C1 controls); the size limit is now the string
length, which is the wire byte count for a ByteString. Failure stays
CREDENTIAL_HEADER_VALUE_INVALID, i.e. AUTH_INTEGRATION_FAILURE before
dispatch.
- checkFinalInvariants: builds `new Headers(headers)` first; any value
Fetch would reject fails as HEADER_VALUE_INVALID (new additive member of
FinalInvariantFailure) with zero fetch calls.
- validateCredentialPatch: rejects case-only duplicate header names.
RED before fix: http-execution-v3-live-authority.test.ts "rejects a NUL /
non-Latin-1 credential value before dispatch" (got TRANSPORT_FAILURE, fetch
called), http-contract-bridge.test.ts "rejects a NUL/non-Latin-1/tab/DEL/C1
control value ..." and "fails before dispatch when Fetch would reject a
header value", auth-session-adapter.test.ts "rejects credential headers
that differ only in case instead of merging them". Pinned: "admits an
ordinary bearer token", CR/LF rejection, existing "Bearer ok"/"Bearer
token" integration paths.
Gates: check:architecture, check:types:app, check:types:test, lint,
test:component, test:integration, test:contract, test:platform,
test:reference-feature pass. test:unit passes except removal-fixture.test.ts
(needs generated artifacts/ release evidence absent in a fresh worktree;
unrelated).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FE-GATE-020 proves a capability can be removed by rebuilding the whole project
without it. The fixture it built could not get that far, and the failures all
came from the fixture rather than from anything about removability.
It was not a repository. The supply-chain inventory is defined as the tracked
file set, so it asks `git ls-files` what the project contains; with no
repository to ask, generation failed and took every provider suite down with
it. It is now initialised on preparation and committed after the removal — not
before, or the index would still list the files the removal deleted.
It had no `.gitignore`, so once it did have a repository, every generated
artifact and every linked module landed in the index and the inventory refused
the fixture for tracked and generated paths colliding. It carries the ignore
rules now, and therefore records the same tracked set as the repository it was
copied from.
Each removal script kept its own copy-target list and they had drifted: the
reference-feature fixture omitted `playwright.capabilities.config.ts`, which the
inventory requires. There is one list now. It also gained the install and
workspace identity — `.npmrc`, the lockfile, the workspace file — without which
the fixture is a different project, and the release evidence a candidate is
assembled from, without which no candidate can be built at all.
A tracked root the removal deletes is no longer required of the result: the
optional-recipe fixture deletes `recipes/`, and the inventory policy demanded
it back. Roots that are gone are pruned from the fixture's policy.
Two smaller causes. A platform integration file asserted the reference feature's
own route ids, so removing the feature left it importing a deleted module —
typecheck, the test run, coverage and the residue scan all failed on that one
misplaced assertion, which now lives in the feature's test tree. And the
canonical exact-count authority was re-imposed on a contract the fixture
deliberately reduces, failing the fixture for the reduction it exists to prove;
`CI_CONTRACT_MODE` already marked those runs and is now honoured by default.
The reference-feature fixture goes from failing before its first assertion to
1,612 passing with one failure, and that one is the live process-tree
observation test already red on the main tree.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three trust boundaries checked a caller's object and then read it again to
use it. Between those two reads an accessor or a Proxy can answer
differently, so the value that passed validation and the value that was
installed were not the same value.
A credential owner's answer was read field by field outside the auth
boundary: a throwing `kind` getter escaped into the transport catch and an
auth outage reached operators as `NETWORK_FAILURE`. Contract composition
validated a contribution and then copied it, so a policy that answered
10,000 to the ceiling check and 999,999 to the copy installed the second
value. The cursor runtime validated its profile once and re-read it on
every page, so raising `maxPages` after construction widened a cap that
had already been checked.
`src/contracts/exact-snapshot.ts` is the one descriptor-based decoder they
now share: every property is read exactly once, an accessor, a symbol, an
inherited or non-enumerable field and a throwing trap all resolve to a
typed failure, and validation runs on the owned copy.
Separately, the `responseBody: NONE` probe awaited a bare `read()`. The
deadline produced a bounded public result while the raw reader kept its
lease, so the body stayed locked and the outer compensator could not
cancel it. The probe now takes the operation lifetime and owns the cancel
and the lock release itself.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Four fixtures linked the installed dependencies into a throwaway root with a
single directory symlink at <fixture>/node_modules, then ran pnpm inside that
root. pnpm does not recognise the modules directory it finds there and purges
it; with CI=true it does so without a prompt. The purge followed the symlink and
deleted the repository's own node_modules mid-run, so a test suite uninstalled
the workspace it was running in. That is what produced the cascading,
file-unrelated failures a full test:unit run reported, and it happened twice
while running the suites for the adapter re-review.
scripts/lib/fixture-node-modules.ts replaces all four sites: node_modules is a
real directory whose entries are individual symlinks, so a recursive delete
unlinks the fixture's own links instead of walking through one link into the
shared tree. Resolution is unchanged.
tests/unit/fixture-node-modules.test.ts performs the exact recursive delete pnpm
performs and asserts the source tree survives, and check:adapter-inventory now
fails on any reintroduction of the directory-symlink form — verified by putting
the old line back and watching the gate reject it.
A full tests/unit + tests/integration run now leaves the dependencies intact.
removal-fixture, supply-chain and security-followup-archive, the three suites
that had to be excluded before, pass in that run.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
LEG-01. AuthSessionPort.recover now takes the request's lifetime context, and
the raw recovery helper returns data only. The sign-out notification moved to
the site that adopts the result, so a recovery that answers after the deadline
or a caller abort is observed and discarded instead of logging the user out of
a request nobody is waiting on.
LEG-02. The V2 client shares V3's credential admission validator instead of
checking the allowed set alone. A bearer profile whose patch omits, empties,
duplicates or corrupts Authorization now fails closed with zero fetches rather
than dispatching an anonymous request under an authenticated profile.
OPT-NET-01. A cursor loader rejection is re-thrown exactly as it is with no
signal at all. Only a signal that has actually aborted classifies the outcome
as PAGINATION_ABORTED, so a real upstream failure stops being filed as a user
cancellation.
OPT-NET-02. defineMutationIntent and the V3 admission site now share the single
isValidIdempotencyKey authority, closing the drift that let a control character
through intent definition.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
LIVE-01. A credential collaborator that returns UNAVAILABLE, throws, rejects
or answers off-contract is an outage of the auth integration, not evidence
about the user's session. Each of those now closes as AUTH_INTEGRATION_FAILURE
with zero fetches, so the composition root's logout path stays reserved for a
genuinely absent session. The synchronous and asynchronous failure sites share
one classifier.
LIVE-02 / LIVE-03. Object.freeze(new Map(...)) freezes the wrapper, not the
backing store, so an exported registry could still be cleared or replaced after
composition. Both the installed REST auth profile registry and the composed
HTTP/event lookups are now read facades over private stores, and every composed
row is an exact own-data snapshot that rejects accessors, inherited and
symbol-keyed fields.
LIVE-04. The total deadline now bounds the physical waits rather than being
checked between them: dispatch and response admission race the attempt signal,
the bounded reader takes that signal, and an abandoned operation is still
observed once so a late native rejection cannot surface unhandled. A body that
completes after the deadline or the caller owns the execution is no longer
admitted; a stale generation keeps its more specific SCOPE_FENCED verdict.
LIVE-05. DEADLINE is no longer treated as a caller-owned cancellation, so a
timeout reaches api.request.failed exactly once while caller, route, scope and
shutdown aborts stay excluded.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
N-06: export one idempotency-key authority from mutation-intent.ts and use it
in the V2 client. A caller-supplied key is validated before credentials, timers
and fetch, and an invalid value is rejected as VALIDATION_REJECTED /
IDEMPOTENCY_KEY_INVALID rather than trimmed, regenerated or dropped, so a keyed
command can no longer replay while sending no key.
N-07: bound the legacy credential wait by the existing attempt controller,
which already carries the total deadline and the caller signal, so a
non-cooperative owner cannot hold the request open and no extra timer is
introduced. The owner receives the operation context, and the failure follows
ownership: deadline to REQUEST_TIMEOUT, caller to REQUEST_ABORTED, and only a
genuine rejection to AUTH_INTEGRATION_FAILURE. None of these paths fetch.
N-08: readBoundedJson delegates to the common bounded reader, so cancel and
releaseLock throws stay isolated inside the closed result, and the V2
content-type mismatch now cancels the response body.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Separate per-attempt physical state from the logical execution history. The
executor now keeps one monotonic certainty accumulator joined through
joinMutationEffectCertainty, records MAYBE_APPLIED at dispatch, and reads the
accumulator from every retry-loop fence, final-invariant, cancellation and
timeout return.
A retry-time scope fence landing between the loop-entry check and the
pre-dispatch invariant can no longer downgrade an already dispatched command to
NOT_STARTED.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Install the REST auth profile registry once at composition and make it the
single transport authority for V3. Contract composition now rejects an
unregistered authProfileId, so the executor never resolves a profile at
runtime.
The credential collaborator contributes proof headers only: Fetch credentials
come from the resolved profile, transport-owned and forbidden headers are
rejected, headers outside the profile's allowed set are rejected, and a missing
required header fails closed as AUTH_INTEGRATION_FAILURE with zero fetch calls.
The final invariant re-proves credentials mode and the exact header sets.
Demo mode satisfies the strict bearer profile with a fixed non-secret marker
instead of weakening REFERENCE_EXTERNAL_BEARER. Credential owners now receive
the operation lifetime through AuthOperationContext.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Project one typed HttpExecutionObservation per logical V3 execution through a
closed composition-root projector: only registered diagnostic context keys and
bucketed values reach the sinks, and terminal non-abort failures now emit
exactly one api.request.failed telemetry event. Caller cancellation and scope
fencing record a diagnostic but never a failure event.
routeId becomes a required input at the installed operation-executor boundary
so the feature gateway's low-cardinality route identity survives to the sink.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>