[psychologister]

Goal And Delivered Outcome

• Outcome basis / optional anchor: Self-contained ordinary maintenance contribution (single-owner convergence); no issue or roadmap id invented.
• Goal/source and gap: loopx/capabilities/issue_fix/repository_context.py owns the closed trust/freshness sets every issue_fix source payload is validated against. reviewer_recommendation.py restated both under its own names (REVIEWER_SOURCE_TRUST_LEVELS, REVIEWER_SOURCE_FRESHNESS_STATES), so one vocabulary decision had two owners and a drift would let the two validators disagree about the same fields.
• Observable before → after, with the validation row that proves it: Before — two modules each define the same two sets, so a renamed or value-drifted copy is undetected. After — reviewer_recommendation.py imports both as aliases of the owner objects, and a new census guard fails if any second definition reappears (see the unit and regression_parity rows).
• Issue/task and intended base: Related to the single-owner convergence direction; base main.

Author Declaration

• Written by: model_agent

Implemented against

• Specification and revision: no written specification; the request in this PR is the basis.
• Criteria:

Criterion (spec clause)	Disposition	Symbol / path	Test or command
The two source vocabularies are stated only by the owner and read by identity elsewhere	implemented	loopx/capabilities/issue_fix/repository_context.py (TRUST_LEVELS, FRESHNESS_STATES); reviewer_recommendation.py aliases	tests/architecture/test_issue_fix_source_vocabulary_owner.py

• Self-check before submission: ran the new guard, the issue_fix capability subset, and the full tests/architecture suite on both a pristine base worktree and the head worktree and compared failure sets; ran three mutations against the guard; ran ruff, the public-boundary scan, and the semantic-vocabulary drift smoke. Deliberately left REVIEWER_SOURCE_KINDS local (narrower than the owner) and did not touch the same freshness triple in reward_memory/candidate_review.py (a different capability).

Scope And Continuation

• Completed scope and remaining work: complete within this scope. The fold removes the duplicate definitions and adds a guard that pins single ownership by value, by identity, and by wiring through both public builders.
• Slice boundary / successor: N/A — the maintenance outcome is complete. The reward_memory/candidate_review.py freshness triple is a separate capability and a possible future convergence, not part of this slice.

Validation

• Tested revision: 682af24
• Run state: finished
• Input classes: synthetic, public_fixture

Check kind	Result	Public-safe evidence / limitation
unit	passed	python -m pytest tests/architecture/test_issue_fix_source_vocabulary_owner.py — 8/8 passed; census-by-value, identity, and wiring through both public entry functions, with positive and negative controls.
regression_parity	passed	Pristine base vs head on identical selections: tests/architecture base 2 failed / 1842 passed, head 2 failed / 1850 passed; issue_fix subset base 2 failed / 359 passed, head 2 failed / 367 passed. Failure sets byte-identical (pre-existing test_content_digest_single_owner, test_posix_flag_portability, and test_issue_fix_maintainer_correction_admission::test_t06_stopped_goal_refuses_at_effect_time[file/sqlite]); the +8 are the new guard. Three mutations each caught: restored literal copy → census + identity; renamed copy → census only; divergent reviewer set (drops advisory) → identity + wiring while census stays green.
static	passed	python -m ruff check clean on both changed files; the new guard is ruff format --check clean. The production file was already format-dirty on base; this change is format-neutral and does not reformat it.
real_entrypoint	passed	loopx check --scan-path <changed files> reports public boundary scan clean: 2 files; the unrelated registry-goal error reproduces identically on the pristine base.
integration	passed	python examples/semantic-vocabulary-drift-smoke.py reports ok with byte-identical output between base and head; the two sets are not registered kernel vocabularies, so no ratcheted budget moves.
static	not_run	mypy is not installed in the local environment; neither changed file is in the pinned mypy scope, so the CI mypy gate is unaffected.

• Coverage and gaps: the changed paths are the two vocabulary definitions and their single consumer (_normalise_reviewer_sources); the guard exercises both public builders across the full owner cross-product and rejects out-of-set values. The full tests/ tree was not run locally — scope was tests/architecture (complete) plus the issue_fix capability subset; the two pre-existing failures are unrelated and reproduce on the pristine base. No UI impact.

Frontend / Visual Evidence

• UI impact: none
• Before: N/A
• After: N/A
• First viewport and preview: N/A
• States and viewports shown: N/A
• Source data: none
• Attention review: N/A
• Craft review: N/A

Type of Change

• Bug fix
• New feature
• Breaking change
• Refactoring (no functional changes)
• Documentation update
• Test update

LoopX Area

• Control plane (goals, todos, quota, scheduler, registry, runtime)
• Benchmark boundary (adapters, runners, verifiers, scoring, evidence)
• Capability or extension (providers, adapters, skills)
• Public docs or presentation surface (README, protocols, dashboard)
• Build, packaging, installer, or CI
• Host or runtime integration

Technical Direction

• Direction / acceptance reference, when applicable: N/A (single-owner maintainability convergence).

Shared-authority RFC fixture impact

• Production-scale fixture schema: N/A
• Semantic dimensions changed, or reviewed no-impact rationale: N/A — no TypeScript control-plane or shared Goal Authority claim.
• Provider conformance arms run: N/A
• Read-only legacy/file/PostgreSQL three-arm rehearsal: N/A

Boundary Checklist

• Neither the diff nor this PR body/comments/attachments disclose private state, credentials, raw traces or verifier output, internal links, or local machine paths.
• I did not duplicate maintainer-owned benchmark work unless a maintainer split out a public issue for it.
• I kept the change scoped to the linked issue/task.
• Existing English/Chinese RFC pairs both have a diff and the local premerge check passes, or no paired RFC changed. (no paired RFC changed)
• I completed the visual evidence section for UI changes, or marked UI impact none.
• Every commit includes a DCO Signed-off-by trailer (git commit -s).

[loopx-agent]

Reviewer: model_agent / gpt-6.1-sol / OpenAI / runtime_reported / reasoning effort max

动机 / Motivation

复审提交 682af24，目标主干 8ade78c。这次改动涉及调整 issue_fix 来源信任级别或时效性取值的维护者，以及走两个入口的调用方。此前改一个允许值要同步两份集合，漏改会让工作规划与评审人推荐对同一来源给出不同判定。现在两个入口读同一对象，基线中第二份声明可由新增测试检出。同一隔离 Git 夹具对照 24 个 CLI 输入。九组合法组合、无来源回退及四项非法取值检查的 JSON 与退出码逐项一致。验证仅涉及本地公开入口与合成仓库。真实外发评审请求及仓库全量测试未重跑。

改动思路 / Approach

工作规划既有 repository_context 定义取值，推荐入口只导入同一对象。两个入口继续验证各自的来源结构。本 PR 只收敛 issue_fix 的 trust 和 freshness 两个词表。工作规划通过既有入口验证 source，推荐侧沿同一集合校验后仍按本地证据生成候选。推荐入口自身只接收 maintainer_map，这个条件继续由原校验器负责。该维护改动以 PR 说明和现有 caller 为验收依据。工作规划的校验归属 · 评审推荐入口

具体改动 / Changes

关键代码讲解

• repository_context 持有 TRUST_LEVELS 与 FRESHNESS_STATES。reviewer_recommendation 将原模块名作为 import alias，原 _normalise_reviewer_sources 继续验证字段和值。CLI 传入原格式，成功响应、失败诊断及候选的外发权限字段都保持一致。允许值 · 导入 · 推荐来源验证
• 新测试从三个角度验证变更。扫描覆盖本包的普通 set 字面量，核对实际 import 对象，再用两个公开 builder 覆盖所有合法值和非法字段。把同一守卫放到基线，可检出原重复和对象分离。只在隔离进程内移除 advisory，真实推荐 builder 的全值用例随即失败。定向测试

对主干的风险 / Risks

本地实际运行 reviewer-plan --execute 和 workflow-plan，共比对 24 行成功、缺省及拒绝结果。核对完整 JSON、退出码和诊断。原计划与推荐的字段、排序和权限投影逐项相同。基线 42 项和提交 50 项定向回归全通过。词汇漂移检查输出相同，ruff、编译、diff 与两文件公开边界检查通过。REST 状态显示分支落后于 main，所触及的生产 owner 和消费者在分叉点与当前 main 对应文件相同。

语义与 CI 对齐

本 PR 复用 issue_fix 包内现有闭集。未定义新内核词汇。有效性证据来自上述本地检查。原词表

我的整体评价 / Verdict

通过当前提交的代码审查。两个入口复用同一校验词表，减少再改取值时需要同步的地方。真实入口的成功、失败与默认回退保持原行为。定向验证覆盖本次变更的输入和值域。验证仅涉及本地公开入口与合成仓库。真实外发评审请求及仓库全量测试未重跑。

English verdict: APPROVE. Both issue_fix callers now use the existing vocabulary owner. All 24 matched CLI outcomes preserve the current behavior.