test(core): consolidate env, telemetry, canonical-origin coverage

.github/workflows/docker-publish.yml

@@ -32,21 +32,18 @@ jobs:
           - component: backend
             image: ai_email_client-backend
             dockerfile: Dockerfile
-            base_dockerfile: Dockerfile
             context: .
             build_args: |
               BUILDKIT_INLINE_CACHE=1
           - component: naruon
             image: naruon
             dockerfile: Dockerfile
-            base_dockerfile: Dockerfile
             context: .
             build_args: |
               BUILDKIT_INLINE_CACHE=1
           - component: frontend
             image: ai_email_client-frontend
             dockerfile: frontend/Dockerfile
-            base_dockerfile: frontend/Dockerfile
             context: .
             build_args: |
               BUILDKIT_INLINE_CACHE=1
@@ -67,28 +64,9 @@ jobs:
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@bb05f3f5519dd87d3ba754cc423b652a5edd6d2c # v4.2.0
 
-      - name: Resolve pinned Ollama base manifest
-        if: matrix.component == 'naruon'
-        run: |
-          base_image="$(awk 'toupper($1) == "FROM" { print $2; exit }' Dockerfile.ollama)"
-          if ! printf '%s\n' "$base_image" | grep -Eq '^ollama/ollama@sha256:[0-9a-f]{64}$'; then
-            printf '::error file=Dockerfile.ollama,line=1::Expected an exact ollama/ollama sha256 base pin; found %s\n' "$base_image"
-            exit 1
-          fi
-          printf 'Resolving pinned Ollama base manifest: %s\n' "$base_image"
-          manifest_output="$(docker buildx imagetools inspect "$base_image")"
-          printf '%s\n' "$manifest_output"
-          for platform in linux/amd64 linux/arm64; do
-            if ! printf '%s\n' "$manifest_output" | grep -Eq "^[[:space:]]*Platform:[[:space:]]+${platform}[[:space:]]*$"; then
-              printf '::error file=Dockerfile.ollama,line=1::Pinned Ollama manifest is missing %s\n' "$platform"
-              exit 1
-            fi
-          done
-
       - name: Prepare OCI annotation values
         id: oci
         env:
-          BASE_DOCKERFILE: ${{ matrix.base_dockerfile }}
           GIT_REF_NAME: ${{ github.ref_name }}
           IMAGE_COMPONENT: ${{ matrix.component }}
           IMAGE_NAME: ${{ matrix.image }}
@@ -98,29 +76,24 @@ jobs:
           version="$(cat VERSION)"
           created="$(date -u +'%Y-%m-%dT%H:%M:%SZ')"
           vendor="${REPOSITORY%%/*}"
-          base_reference="$(awk 'toupper($1) == "FROM" { print $2; exit }' "$BASE_DOCKERFILE")"
-          if ! printf '%s\n' "$base_reference" | grep -Eq '^[A-Za-z0-9._/-]+:[A-Za-z0-9._-]+@sha256:[0-9a-f]{64}$'; then
-            printf '::error file=%s,line=1::Expected an exact tagged sha256 base pin; found %s\n' "$BASE_DOCKERFILE" "$base_reference"
-            exit 1
-          fi
-          base_digest="${base_reference##*@}"
-          base_repository="${base_reference%@*}"
-          case "$base_repository" in
-            */*) base_name="$base_reference" ;;
-            *) base_name="docker.io/library/$base_reference" ;;
-          esac
           case "$IMAGE_COMPONENT" in
             frontend)
               title="naruon frontend"
               description="Naruon Next.js frontend runtime image"
+              base_digest="sha256:191ef878ecb351d68b78219593de18bd8942afd59af59f29960dc4b24805a3f1"
+              base_name="docker.io/library/node:26-slim@${base_digest}"
               ;;
             backend)
               title="naruon backend"
               description="Naruon FastAPI backend runtime image"
+              base_digest="sha256:44dd04494ee8f3b538294360e7c4b3acb87c8268e4d0a4828a6500b1eff50061"
+              base_name="docker.io/library/python:3.14-slim@${base_digest}"
               ;;
             *)
               title="naruon"
               description="Naruon combined FastAPI and Next.js runtime image"
+              base_digest="sha256:44dd04494ee8f3b538294360e7c4b3acb87c8268e4d0a4828a6500b1eff50061"
+              base_name="docker.io/library/python:3.14-slim@${base_digest}"
               ;;
           esac
           {
@@ -185,21 +158,18 @@ jobs:
           - component: backend
             image: ai_email_client-backend
             dockerfile: Dockerfile
-            base_dockerfile: Dockerfile
             context: .
             build_args: |
               BUILDKIT_INLINE_CACHE=1
           - component: naruon
             image: naruon
             dockerfile: Dockerfile
-            base_dockerfile: Dockerfile
             context: .
             build_args: |
               BUILDKIT_INLINE_CACHE=1
           - component: frontend
             image: ai_email_client-frontend
             dockerfile: frontend/Dockerfile
-            base_dockerfile: frontend/Dockerfile
             context: .
             build_args: |
               BUILDKIT_INLINE_CACHE=1
@@ -234,7 +204,6 @@ jobs:
       - name: Prepare OCI annotation values
         id: oci
         env:
-          BASE_DOCKERFILE: ${{ matrix.base_dockerfile }}
           GIT_REF_NAME: ${{ github.ref_name }}
           IMAGE_COMPONENT: ${{ matrix.component }}
           IMAGE_NAME: ${{ matrix.image }}
@@ -245,29 +214,24 @@ jobs:
           version="${VERSION_VALUE:-$(cat VERSION)}"
           created="$(date -u +'%Y-%m-%dT%H:%M:%SZ')"
           vendor="${REPOSITORY%%/*}"
-          base_reference="$(awk 'toupper($1) == "FROM" { print $2; exit }' "$BASE_DOCKERFILE")"
-          if ! printf '%s\n' "$base_reference" | grep -Eq '^[A-Za-z0-9._/-]+:[A-Za-z0-9._-]+@sha256:[0-9a-f]{64}$'; then
-            printf '::error file=%s,line=1::Expected an exact tagged sha256 base pin; found %s\n' "$BASE_DOCKERFILE" "$base_reference"
-            exit 1
-          fi
-          base_digest="${base_reference##*@}"
-          base_repository="${base_reference%@*}"
-          case "$base_repository" in
-            */*) base_name="$base_reference" ;;
-            *) base_name="docker.io/library/$base_reference" ;;
-          esac
           case "$IMAGE_COMPONENT" in
             frontend)
               title="naruon frontend"
               description="Naruon Next.js frontend runtime image"
+              base_digest="sha256:191ef878ecb351d68b78219593de18bd8942afd59af59f29960dc4b24805a3f1"
+              base_name="docker.io/library/node:26-slim@${base_digest}"
               ;;
             backend)
               title="naruon backend"
               description="Naruon FastAPI backend runtime image"
+              base_digest="sha256:44dd04494ee8f3b538294360e7c4b3acb87c8268e4d0a4828a6500b1eff50061"
+              base_name="docker.io/library/python:3.14-slim@${base_digest}"
               ;;
             *)
               title="naruon"
               description="Naruon combined FastAPI and Next.js runtime image"
+              base_digest="sha256:44dd04494ee8f3b538294360e7c4b3acb87c8268e4d0a4828a6500b1eff50061"
+              base_name="docker.io/library/python:3.14-slim@${base_digest}"
               ;;
           esac
           {

.jules/bolt.md

@@ -19,10 +19,3 @@
 
 **Learning:** When using a dictionary purely to track the presence of keys (e.g. `has_sent_message[key] = True`), checking for presence with `.get(key, False)` carries unnecessary semantic and memory overhead. Sets in Python provide a cleaner `key in set_name` syntax for boolean presence checks and slightly reduced memory footprint, while maintaining O(1) time complexity.
 **Action:** When tracking unique occurrences or boolean presence of items where the value itself doesn't carry additional information, use a `set` and its `.add()` and `in` operators instead of a `dict` mapping to `True` or `False`.
-## 2025-02-12 - Replaced O(N) Array Lookups with O(1) Maps in Loops
-
-**Learning:** When generating derived UI state in `useMemo` that joins separate data arrays (like graph edges referencing node IDs), calling helper functions that use `Array.prototype.find()` for every item creates an `O(M * N)` bottleneck.
-**Action:** When a loop needs to repeatedly look up related items from another array by ID, pre-compute an `O(N)` `Map` before the loop and use `map.get()` for `O(1)` lookups instead of inline array `.find()` calls.
-## 2024-05-24 - [React Component Memoization]
-**Learning:** In React components like `WorkspaceHome`, when layout state or polling changes trigger parent re-renders, expensive child components like `EmailDetail` will also re-render unnecessarily if not memoized.
-**Action:** Always consider `React.memo` for heavy child components that rely on stable props (like IDs) when the parent component has frequent unrelated state updates.

.jules/sentinel.md

@@ -129,7 +129,3 @@
 **Vulnerability:** The URL validation logic correctly blocked non-global IP addresses and `localhost`, but failed to block internal domain extensions such as `.internal` or `.local` (or exact matches for `internal`). This could allow attackers to bypass SSRF protections by resolving these internal top-level domains.
 **Learning:** Checking for `localhost` alone is insufficient to prevent SSRF against internal network resources, as modern environments and protocols utilize `.internal` and `.local` domains for internal routing.
 **Prevention:** Always explicitly check and block domains matching `.internal`, `.local`, or `internal` (alongside `localhost`) when validating URLs for global reachability to prevent SSRF bypasses.
-## 2025-02-23 - CRLF Injection in Email Headers
-**Vulnerability:** The `in_reply_to` and `references` fields on the `SendEmailRequest` model lacked explicit validation, opening up an opportunity for header injection by appending `\r\n`.
-**Learning:** While the email service internally checks some headers, relying on the API boundary's Pydantic model ensures bad input is stopped early and consistently. Pydantic regex patterns aren't sufficient on their own for all string contexts due to encoding/decoding inconsistencies.
-**Prevention:** Always use `@field_validator` with explicit `mode="before"` string matching to reject `chr(10)` and `chr(13)` across all user-controlled email header fields. Use `isinstance(value, str)` before string operations to prevent runtime errors if input is missing or malformed.

.trivyignore

@@ -0,0 +1 @@
+CVE-2026-67213

AGENTS.md

@@ -95,17 +95,6 @@ in this repo.
   knowledge-graph pipeline (DOM decomposition, entity/relation extraction,
   grounded graph retrieval) should ground itself in the relevant layout-analysis
   and knowledge-graph / grounded-retrieval literature.
-
-### Structural topic-model boundary
-
-- Do not implement or describe hard-coded term lists, term frequency,
-  embeddings, or LLM-assigned labels as structural topic modeling (STM).
-  Fixed business labels are not topic-posterior estimates, and the explicitly
-  lexical `keyword_extractor` must not be used as topic evidence.
-- Topic inference requires a versioned fitted TEPP model and its frozen
-  preprocessing and vocabulary contract. If that fitted model is unavailable,
-  fail closed; do not return a default label, template agenda, or substitute
-  keyword/embedding/LLM result presented as STM.
 <!-- END cwl-agent-guidance -->
 
 ## Release governance defaults
@@ -436,15 +425,6 @@ in this repo.
 - Public audit/event identifiers that may use human-readable prefixes must not
   be stored in artificially short `varchar(n)` columns; use opaque source UIDs
   that fit seeded smoke data and provider evidence without truncation.
-- Conceptual ERDs, API schemas, persistence models, and fixtures must not mark a
-  reusable business identifier such as `document_ref`, `model_id`, `topic_id`,
-  or `label_id` as an unscoped primary or foreign key. Use an opaque immutable
-  reference that binds the full scope or an explicit composite identity with the
-  applicable snapshot revision, model version, request/result scope, or label
-  version. Define the required identity tuple for each entity; require only the
-  dimensions relevant to that entity. Never join snapshots, model artifacts,
-  topic components, or label evidence by a bare document, model, topic, rank,
-  label, or display value.
 - When reviews find public/private identifier leaks, stale API fixture shapes, or recurring bug patterns, update tests, frontend mocks, E2E mocks, README examples, architecture docs, and explicitly record the anti-pattern in `AGENTS.md` so the same bug pattern does not reappear in copied examples.
 - When reviews find missing browser security headers or tabnabbing hardening,
   update both backend header tests and frontend link tests. Global backend

ARCHITECTURE.md

@@ -20,21 +20,6 @@ Runtime database connectivity is secret-injected: `backend/core/config.py` has
 no fallback `DATABASE_URL`, so missing database configuration fails at startup
 rather than silently using shared development credentials.
 
-## Topic-intelligence boundary
-
-Naruon has no live Structural Topic Modeling endpoint, fitted topic artifact,
-or topic-result persistence. The retained `keyword_extractor` is deterministic
-lexical metadata and must not feed topic, agenda, search, or norm-group
-inference. A future adapter may consume a separately accepted, versioned TEPP
-artifact/API only when frozen preprocessing and vocabulary, covariate design,
-mixed-membership posterior uncertainty, diagnostics, provenance, and explicit
-abstention are all available. Missing or incompatible scientific authority
-fails closed. Naruon owns authentication, authorization, request validation,
-the adapter envelope, and disclosure policy; TEPP would own the scientific
-payload. See the canonical documentation graph in
-[`docs/topic-intelligence/README.md`](docs/topic-intelligence/README.md) and
-[`ADR-0001`](docs/adr/0001-topic-measurement-authority.md).
-
 ## Workspace navigation boundary
 
 The Next.js shell opens the Today execution dashboard for first-run sessions and

CHANGELOG.md

@@ -1,28 +1,11 @@
 ## [Unreleased]
-- EmailDetail 테스트가 지원하지 않는 스레드 병합/분리 버튼을 `textContent`뿐 아니라 `aria-label`과 `title` 접근 가능 이름으로도 검출하도록 바꿔, 아이콘 전용 버튼 회귀를 놓치지 않습니다.
-### 주제 측정 경계 (Topic Measurement)
-
-- STM 결과로 오인될 수 있었던 하드코딩 용어표 기반
-  `email_categorizer`와 `meeting_agenda_generator`를 도구 레지스트리에서
-  제거했습니다. `keyword_extractor`는 결정론적 단어 빈도 유틸리티로 유지하되
-  주제 posterior 근거로 사용하지 않는 경계를 문서화했습니다. 현재 Naruon에는
-  fitted TEPP 모델 기반 production 주제 측정 API가 없으므로, 모델 부재 시
-  기본 라벨이나 템플릿으로 대체하지 않고 fail closed 합니다.
-- 이 경계의 PRD, TRD, ADR, Architecture, API 계약, JSON Schema, UML,
-  개념 ERD, 보안·위협 모델, 테스트·운영 전략, 추적성 및 문서 적합성 평가를
-  `docs/topic-intelligence/`에 하나의 상태 표시 문서 그래프로 정리했습니다.
-  이는 미래 계약의 설계 근거이며, 현재 runtime 구현이나 물리 DB 엔터티가
-  존재한다는 주장이 아닙니다.
-- UUID V4 제너레이터(`uuid_v4_generator`) 도구를 추가하여 런타임에서 범용 고유 식별자 버전 4를 랜덤으로 생성할 수 있게 하였습니다. 테스트 커버리지 100%를 보장합니다.
-
 ### 보안 패치 (CodeQL extended current-head)
 
 - `cryptography`를 `50.0.0`으로 갱신해 공격자 제공 PKCS#7 EnvelopedData 복호화 결과의 오류·타이밍 차이로 발생하는 Bleichenbacher oracle(`CVE-2026-69247`, `GHSA-g6cj-pr64-35w5`)을 제거하고, backend·uv lock·hash lock·Strix CI 의존성 증거를 같은 버전으로 동기화했습니다. Strix 잠금은 `google-cloud-aiplatform==1.160.0`의 `<7` 제약을 위반하던 `protobuf==7.35.1`을 이미 검증된 `6.33.6`으로 복구해 다시 해석·설치 가능하게 했습니다.
 - CodeQL `extended` 기본 설정이 current `develop`에서 확인한 Critical 8건·High 21건·Medium 1건을 코드 경계에서 제거합니다. 서버 요청은 검증된 loopback/HTTPS origin, 동일 OIDC issuer origin, 허용 API 경로·쿼리만 재구성하고 redirect를 자동 추종하지 않으며, 공개 IPv6 authority를 보존합니다. UI smoke는 고정 Node/Next 실행 파일과 인자, localhost:3001 allowlist, private `mkdtemp` artifact 디렉터리 및 containment 검사만 사용합니다.
 - OIDC token endpoint는 운영 환경에서 서버 전용 `OIDC_ALLOWED_HOSTS` 정확 호스트 allowlist를 필수로 적용합니다. hostname의 모든 DNS 결과가 공인 주소인지 검증한 뒤 해당 주소 집합을 native HTTP(S) 연결의 `lookup`에 고정하고, 원래 issuer hostname은 Host/TLS SNI로 유지해 사설 주소 해석과 DNS rebinding 사이의 TOCTOU를 차단합니다. 실패 로그는 입력 URL·token 대신 고정된 configuration/DNS·transport/response/backend-verification reason code만 남깁니다.
 - Trivy 2026-07-26 DB에서 새로 확인된 Next.js High 4건·Medium 5건(`CVE-2026-64641`–`CVE-2026-64649`)과 PostCSS High 1건(`GHSA-r28c-9q8g-f849`)을 제거하기 위해 Next.js/`eslint-config-next`를 `16.2.11`, PostCSS를 `8.5.18`로 갱신했습니다. 이후 2026-08-04 DB가 `8.5.18`에서 추가 탐지한 PostCSS Medium(`CVE-2026-69153`, 최초 수정 `8.5.23`)도 제거하도록 manifest·workspace override·lock을 `8.5.24`로 동기화했으며 저장소의 release-age 정책을 우회하지 않습니다.
 - `pnpm audit`가 개발 도구 체인에서 추가 탐지한 `brace-expansion <=5.0.7` High DoS(`GHSA-mh99-v99m-4gvg`)와 이후 `5.0.8`까지 영향을 주는 우회형 High DoS(`GHSA-rgw5-rvv9-x895`)는 `5.0.9` 전역 override로 제거했습니다. CommonJS default export를 기대하는 legacy `minimatch 3.1.5`에는 `expand` named export도 수용하는 최소 pnpm 패치를 적용해 ESLint/glob 동작을 보존합니다. 같은 감사에서 확인된 `undici 7.28.0`의 High 1건·Moderate 4건(`GHSA-4cwx-7wf7-3272` 등)은 `jsdom 30.0.1` 및 release-age 정책을 통과하는 `undici 8.9.0`으로 갱신했습니다.
-- PostCSS의 Nano ID 해석을 `3.3.18`로 갱신해 사용자 제공 음수 크기에서 비보안 생성기가 무한 반복될 수 있는 High DoS(`CVE-2026-67214`, `GHSA-28wg-ghj8-5hjv`)를 제거했습니다. lockfile과 release-governance 회귀 테스트가 같은 최초 수정 3.x 버전을 강제합니다.
 - root·frontend Docker build의 frozen install 계층이 pnpm manifest와 함께 `frontend/patches`를 먼저 복사하도록 수정해, 이미지 검증에서도 lockfile의 patched dependency를 동일하게 재현합니다.
 - Scorecard SARIF normalizer는 고정 workspace artifact로 정규화되는 `./scorecard-results.sarif`와 절대 경로를 동일하게 허용하면서 symlink·workspace 이탈은 계속 거부합니다. 도구 실행 실패 API는 CR/LF·제어 문자를 escape하고 500자로 제한하며, 로그에는 raw 도구 코드·예외 text 대신 SHA-256 기반 코드·traceback 상관 식별자만 기록합니다.
 - 백엔드 origin 보안 경계를 `frontend/src/lib/backend-url.ts`의 단일 생성기로 통합해 API proxy·session·OIDC callback이 같은 검증을 사용합니다. UI smoke의 새 `NARUON_FULL_PRODUCT_SCREENSHOT_PROFILE` 이름은 실제 selector 의미를 드러내며, 기존 `..._SCREENSHOT_DIR`은 호환 alias로 계속 지원합니다.

CLAUDE.md

@@ -146,13 +146,6 @@ Next.js frontend ──> FastAPI backend (control plane) ──> Postgres + pgve
   test/lint/build), plus `bandit`, `codeql`, `trivy`, `scorecard`,
   `pr-governance`, `docker-publish` (GHCR on `v*` tags matching `VERSION`), and
   `mail-smoke`. Actions are pinned to full commit SHAs.
-- Topic intelligence is **not implemented**. Never use lexical frequencies,
-  embeddings, zero-shot labels, or request-time LLM labels as an STM result.
-  The retained `keyword_extractor` is lexical metadata only. Any future adapter
-  is blocked on a versioned fitted TEPP artifact/API with frozen preprocessing,
-  mixed-membership uncertainty and diagnostics; absence or incompatibility
-  fails closed. Start at `docs/topic-intelligence/README.md` and
-  `docs/adr/0001-topic-measurement-authority.md`.
 
 ## Key conventions