feat(ask): nominate persisted semantic evidence

[seonghobae]

Product gap

Global Ask loaded persisted project, role, person, organization, team, and Knowledge Graph evidence only after embedding candidate selection. A term present only in that evidence could miss its authorized source (issue #272, internal nomination slice).

Change

• nominate bounded IDs with PostgreSQL websearch_to_tsquery('simple', ...) and matching expression GIN indexes;
• apply corporate-entity/process-unit scope, publication eligibility, and event-time bounds before the candidate limit;
• repeat the existing authorization boundary before reading source bodies or semantic evidence;
• keep embedding optional and introduce no local score, threshold, tokenizer, or channel weight;
• record ADR 0225, APA 7 research register, changelog, and product-gap evidence.

External SearXNG verification and three-way public truth status remain explicitly outside this slice.

Verification

• uv run --extra dev --extra backend pytest tests/test_global_ask_semantic_candidates.py tests/test_global_ask_sources.py tests/test_public_docstrings.py tests/test_documentation_hygiene.py -q (26 passed)
• migration 0222 applied twice against temporary PostgreSQL; seven GIN indexes remained present
• git diff --check

Closes no issue; partially addresses #272.

---

Summary by CodeRabbit

• 새로운 기능

  • Global Ask가 저장된 프로젝트·역할·인물·조직·팀 및 Knowledge Graph 근거를 활용해 관련 후보를 찾습니다.
  • 지식 기준 시점을 지정해 과거 시점 기준으로 답변과 근거를 확인할 수 있습니다.
  • 선택적으로 공개 정보와 인용된 주장을 외부 검증할 수 있습니다.
  • 답변에 근거 상태, 검증 결과, 제한사항 및 후속 조치가 표시됩니다.

• 개선 사항

  • 검색 전후 권한, 가시성 및 자격 검사를 강화했습니다.
  • 임베딩 검색이 unavailable 상태여도 다른 검색 경로를 계속 사용할 수 있습니다.
  • 인용된 개정본과 이후 변경 여부가 더 명확히 표시됩니다.

• 문서

  • 의미 기반 후보 검색과 공개 주장 검증 관련 안내 및 요구사항을 추가했습니다.

[coderabbitai]

Warning

Review limit reached

Next included review available in 51 minutes.

View limit details

Limit details: You’ve used the included review currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 230330d6-7709-4258-9c4b-fe1b2998220f

📥 Commits

Reviewing files that changed from the base of the PR and between 5d2626b and a3e87a8.

📒 Files selected for processing (5)

• docs/adr/README.md
• docs/product-technical-gap-baseline.md
• docs/storybook-inventory.md
• lineageweave/claim_verification.py
• tests/test_server_diagnostics.py

📝 Walkthrough

Walkthrough

Global Ask가 영속 의미·그래프 근거에서 후보 게시물을 선정합니다. 요청은 knowledge_cutoff를 저장하고 권한과 시간 조건을 적용합니다. 공개 클레임은 선택적으로 외부 검증하며, 프런트엔드는 근거 상태와 제한사항을 표시합니다.

Changes

Global Ask 검색 및 검증

Layer / File(s)	Summary
의미 후보 선정과 검색 인덱스 backend/app/global_ask_semantic_candidates.py, backend/app/post_chat_ingestion.py, migrations/0225_global_ask_semantic_candidate_search.sql, tests/test_global_ask_semantic_candidates.py, tests/test_global_ask_sources.py	PostgreSQL 전문 검색이 영속 프로젝트·역할·인물·조직·팀·Knowledge Graph 근거에서 제한된 후보 ID를 생성합니다. 권한, 상태, 날짜, 자격 조건을 적용합니다. 후보를 임베딩 결과와 결합합니다.
공개 근거 투영 backend/app/post_chat_ingestion.py, tests/test_post_chat.py	그래프 사실을 증거 게시물별로 귀속합니다. 공개 프로젝트와 그래프 관계를 typed public_claims로 변환합니다. 숨겨진 증거와 역사적 제한을 소스에서 제외합니다.
작업 큐와 knowledge cutoff backend/app/global_ask_queue.py, backend/app/main.py, backend/tests/test_api.py, tests/test_global_ask_queue.py, tests/test_server_diagnostics.py	Global Ask 요청과 워커가 knowledge_cutoff와 semantic query factory를 전달합니다. cutoff 조회는 질문 임베딩, 계보 확장, 이미지 수집을 조건부로 건너뜁니다. 입력 검증과 임베딩 실패 처리를 테스트합니다.
공개 주장 검증과 결과 표시 lineageweave/claim_verification.py, tests/test_claim_verification.py, frontend/src/App.tsx, frontend/src/i18n.ts, frontend/src/App.test.tsx	공개 클레임을 정규화하고 verify 모드로 외부 검증합니다. 외부 URL과 내부 게시물 ID를 분리합니다. 프런트엔드는 검증 상태, cutoff 근거, 제한사항, 보존된 개정본을 표시합니다.
계약 및 지원 문서 docs/adr/0233-global-ask-semantic-candidate-nomination.md, docs/adr/0234-global-ask-public-claim-verification.md, docs/product-requirements.md, docs/doctoring/*, docs/storybook-inventory.md, CHANGELOG.md, docs/adr/README.md, docs/product-technical-gap-baseline.md	의미 후보 선정, 공개 검증, knowledge cutoff 및 관련 검증 절차의 계약과 추적 문서를 갱신합니다.

Estimated code review effort: 5 (Critical) | ~120 minutes

Merge Risk: 🔵 Low · up to 5d262

The change expands public claim verification, but configured search-host self-references may currently pass through and create unintended external egress or citation behavior. The PR is otherwise mergeable with explicit owner awareness and follow-up for this bounded security concern, plus minor documentation cleanup.

Sequence Diagram(s)

sequenceDiagram
  participant Client as AskAgentPanel
  participant API as submit_global_ask
  participant Queue as global_ask_queue
  participant Candidates as semantic_candidate_post_ids
  participant Sources as gather_global_chat_sources
  participant Verifier as ClaimVerificationClient
  participant Orchestrator as contextual-orchestrator
  Client->>API: 질문, knowledge_cutoff, 검증 동의 제출
  API->>Queue: 작업 저장 및 워커 전달
  Queue->>Candidates: 질문과 권한 범위 전달
  Candidates-->>Sources: 제한된 후보 post ID 반환
  Sources-->>Queue: 내부 소스와 public_claims 반환
  Queue->>Verifier: 검증 동의 시 공개 클레임 전달
  Verifier->>Orchestrator: verify 모드 판정 요청
  Orchestrator-->>Verifier: 검증 상태 반환
  Verifier-->>Queue: 외부 검증 결과 반환
  Queue-->>Client: 답변, 인용, grounding 상태, 제한사항 반환

Loading

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 47.14% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 70 functions across 21 files. (5 skipped:…	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	제목은 Global Ask에 영속 의미 증거를 후보로 선정하는 이번 변경의 핵심을 간결하고 명확하게 설명합니다.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 47.14% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 70 functions across 21 files. (5 skipped: 5 unsupported.)

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/global-ask-semantic-public-current

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[seonghobae]

Reviewed against ADR 0225 and the shared post_chat_ingestion merge contract.

• The recency ordering is intentional for the persisted semantic nomination slice: this channel is a deterministic, provenance-preserving candidate source, not a relevance score. Rebalancing it against embeddings would introduce an unsupported weighting rule; downstream embedding/LLM channels remain separate and the final bounded set is explicitly labeled as candidate evidence.
• The iterable materialization change is covered by the generator regression test and is accepted as-is.

No code change is warranted for these informational findings; the contract and test evidence are recorded here.

[seonghobae]

Additional exact implementation evidence:

• uv run --extra dev --extra backend pytest tests/test_static_sql_review_contracts.py tests/test_migration_replay.py tests/test_optional_extra_collection.py tests/test_global_ask_semantic_candidates.py tests/test_global_ask_sources.py -q → 56 passed.
• Migration 0222 applied twice against temporary PostgreSQL and retained exactly seven expression GIN indexes.
• The complete nomination SQL executed through asyncpg against PostgreSQL with one synthetic public project-only match and returned only its expected post UUID.

No production record, organization name, or runtime credential was used or recorded.

[seonghobae]

@coderabbitai review

[coderabbitai]

⚠️ Action not completed

Review rate limited.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[seonghobae]

Applied the code-quality review at exact head 3d743ff7: pass _claim_verification_client directly as the worker factory instead of wrapping it in a redundant lambda. Module compilation passes.