[hhyykk]

Goal And Delivered Outcome

• Outcome basis / optional anchor: Self-contained control-plane contract hardening under CONTRIBUTING.md#find-work and the Kernel/capability responsibility boundary in docs/architecture.md#runtime-responsibility-model.
• Goal/source and gap: bootstrap_project previously recorded an explicit --goal-doc in authority_sources without a content digest lock, did not discover top-level repository specifications (architecture.md, goal-and-requirements.md), and check_contract did not verify whether registered authority sources were deleted or mutated after bootstrap or whether snapshotted repository-wide invariants were violated during replans.
• Observable before → after, with the validation row that proves it:
  • Before: Mutating or deleting a registered authority source or violating declared repository-wide invariants (forbid_import, forbid_pattern, must_exist, immutable_path) passed check_contract (ok=True, errors=[]).
  • After: bootstrap_project records sha256 locks for existing --goal-doc files and discovered architecture.md / goal-and-requirements.md specifications (and renders ## Repo-Wide Architecture Invariants into ACTIVE_GOAL_STATE.md). check_contract runs domain-neutral verify_authority_source_digests unconditionally and runs verify_executable_architecture_invariants when include_architecture_invariants=True (proven by tests/test_architecture_invariants.py, 13 passed).
• Issue/task and intended base: Targets main (self-contained contract increment; no pre-existing issue ID claimed).

Author Declaration

• Written by: model_agent (OpenAI GPT-6.1-sol via Codex CLI and Google Gemini pair-programming assistant, operated by @hhyykk).

Implemented against

• Specification and revision: no written specification; the request in this PR is the basis, aligned with docs/architecture.md at 59464f9741fe177381743eadaaffebbb54e5472e.
• Criteria:

Criterion (spec clause)	Disposition	Symbol / path	Test or command
Lock existing --goal-doc and discovered top-level specs with SHA-256 at bootstrap while allowing non-existent --goal-doc paths to register without a digest	implemented	build_goal_entry in loopx/bootstrap.py; discover_top_level_specs in loopx/control_plane/goals/architecture_invariants.py	pytest -q tests/test_architecture_invariants.py
Separate domain-neutral authority digest verification from optional executable invariant probing	implemented	verify_authority_source_digests, verify_executable_architecture_invariants in loopx/control_plane/goals/architecture_invariants.py; check_contract in loopx/contract.py	pytest -q tests/test_architecture_invariants.py
Enforce forbid_import, forbid_pattern, must_exist, and immutable_path from bootstrap-snapshotted rules without allowing active-state prose to forge lock approvals	implemented	verify_executable_architecture_invariants in loopx/control_plane/goals/architecture_invariants.py	pytest -q tests/test_architecture_invariants.py

• Self-check before submission: Verified 13 focused unit/integration tests in tests/test_architecture_invariants.py, 316 existing bootstrap/contract tests (tests/test_*bootstrap*.py and tests/test_*contract*.py), 298 architecture boundary & content-digest single-owner tests, scripts/generate_semantic_inventory.py --changed-from HEAD, scan_public_boundary, ruff check, and git diff --check. Left full cross-platform CI matrix and live benchmark suites to GitHub Actions CI.

Scope And Continuation

• Completed scope and remaining work: Completed Slice 1 (domain-neutral authority_sources SHA-256 digest locking and verification) and Slice 2 (opt-in include_architecture_invariants probe for forbid_import, forbid_pattern, must_exist, and immutable_path rules snapshotted at bootstrap).
• Slice boundary / successor: Complete within this scope; if maintainers prefer relocating Slice 2's AST/regex invariant probe into a standalone engineering capability package or CLI subcommand, verify_executable_architecture_invariants is already cleanly isolated from verify_authority_source_digests.

Validation

• Tested revision: 576af8b44def7095aaec839c5e159ed49829d88d
• Run state: finished
• Input classes: synthetic, public_fixture

Check kind	Result	Public-safe evidence / limitation
unit	passed	python3 -m pytest -q tests/test_architecture_invariants.py (13 passed: covers explicit/missing --goal-doc digest locking, include_architecture_invariants=False/True, forbid_import, forbid_pattern, must_exist, immutable_path, path traversal rejection, and spec tampering recovery).
regression_parity	passed	python3 -m pytest -q tests/test_*bootstrap*.py tests/test_*contract*.py (303 passed, 1 skipped on existing bootstrap/contract suites with zero regressions; A/B comparison against baseline 59464f9741fe177381743eadaaffebbb54e5472e confirmed baseline missed layer-bypass imports, deleted required modules, and spec rule erasure while candidate caught all 3 invariant violations + unapproved spec mutation).
static	passed	python3 -m pytest -q tests/architecture/test_content_digest_production_owner.py tests/architecture/test_content_digest_single_owner.py tests/architecture/test_control_plane_import_boundaries.py (298 passed); python3 scripts/generate_semantic_inventory.py --changed-from 59464f9741fe177381743eadaaffebbb54e5472e (0 candidates); ruff check on added modules and git diff --check passed; scan_public_boundary passed with 0 hits/warnings.

• Coverage and gaps: The focused unit and regression suites cover bootstrap_project, build_goal_entry, render_state_markdown, check_contract, verify_authority_source_digests, and verify_executable_architecture_invariants. Full repository pytest shards and Windows PowerShell checks were not run locally and will run in PR CI.

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: Core control-plane hardening (authority_sources digest integrity and repository-wide invariant diagnostics).

Shared-authority RFC fixture impact

• Production-scale fixture schema: N/A
• Semantic dimensions changed, or reviewed no-impact rationale: N/A
• Provider conformance arms run: N/A
• Read-only legacy/file/PostgreSQL three-arm rehearsal (required for promotion, runtime-routing, or compatibility-projection changes): 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 (including .loopx/, .codex/goals/, and live ACTIVE_GOAL_STATE.md).
• 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 (including ledger entries) both have a diff, and the local premerge check python3 scripts/check_rfc_bilingual.py --base origin/main --head HEAD passes, or 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

评审对象：576af8b44def7095aaec839c5e159ed49829d88d。两项阻塞：

• [P2] 将软件扫描从默认通用健康路径隔离。 check_contract 的新参数默认是 True，普通 CLI 和 status 调用都不传它。一次性项目仅包含声明缺少文件的规范、没有任何显式启用操作，同输入旧基线 check 为 exit 0，当前为 exit 1；当前直接传 False 又通过。bootstrap 同时已经快照软件规则并生成 Agent 义务，所以内部 bool 也不能隔离完整行为。将源码规则从普通 check/status 的默认路径分离，通过显式、带作用域的工程检查启用；只改 Python 默认参数仍不足以隔离 bootstrap 已生成的新增义务。保留来源校验的独立 owner，并补开/关与旧版默认行为对照。
• [P2] 检查 from-import 中的实际子模块。 ImportFrom 分支 只比较 node.module。声明禁止 pkg.db 时，import pkg.db 会报错，改成 from pkg import db 却返回 ok=True/errors=[]，实际 Python 导入仍加载同一个被禁止的模块。检查 ImportFrom 的模块与导入成员组合，正确处理子模块和相对导入；把 import pkg.db 与 from pkg import db 作为同一禁止依赖的成对反例，避免靠改写合法 import 绕过。

动机

把已有项目接入 LoopX、随后正常检查状态或继续修改代码的用户，会受到自动发现的仓库规范影响。旧版普通 bootstrap 后的 check 不执行这些软件规则；新版仅凭规范文件名就把源码规则接入默认健康检查，即使用户没有显式启用工程扫描。来源摘要锁能发现修改和删除，但当前 head 的默认 CLI 出现新增拒绝，同时一个普通静态导入写法能让禁止依赖检查返回干净结果。本次评审不否定来源完整性需求，也不要求完成整个 TypeScript 迁移；要求把通用完整性与可选软件检查分清，并让已声明规则可靠。

改动思路

来源完整性与软件源代码约束属于两项不同责任；后者不能仅因发现文件名就在通用健康路径默认启用。保留有用的来源校验，先把软件扫描移到显式工程检查边界，并修复已声明导入规则的漏检；无需等待整套控制面迁移。这里应区分三类机制：文件字节及历史接受版本的完整性是通用来源事实；AST/正则扫描是软件工程领域策略；文件读取是只读 provider IO。复用通用诊断容器不会让领域策略自动成为 Kernel 的默认责任。既有注册来源 witness 核的是注册字节身份，与单份文档内容锁不同，不能因同用 SHA-256 就硬合并；已有工程检查能力的显式作用域则是可复用边界。

具体改动

采用作者引用的 docs/architecture.md，固定规格版本 59464f9741fe177381743eadaaffebbb54e5472e：Runtime Responsibility Model 的 capability/domain-policy 与 Kernel 分工未满足；Guided Autonomy, Not Recommendation Lock-In 对显式 typed 强制边界的要求也未满足。修复针对上述当前默认/范围问题，不将未来整套迁移当作前置门。来源摘要在现有测试里能够发现改动、删除，并且 active-state prose 不能批准换锁，这些正向价值保留。

关键代码讲解

1. build_goal_entry 给已存在的显式 goal 文档加摘要，并自动发现四个候选规范位置；它还保存解析规则、原理和 immutable 文件摘要。缺少的显式文档仍保持旧的无摘要注册。
2. check_contract 始终加入来源摘要诊断，并在新增参数为真时加入软件规则。默认值为真且现有公共调用方省略参数，实际行为与“Slice 2 opt-in”不符；bootstrap 的新增指令也没有同一个开关。
3. verify_authority_source_digests 保留旧无摘要记录，对缺少、不可读和内容改变产生 Goal 诊断。需要保留该兼容意义，而不是删除旧来源格式；测试里直接改字典摘要并不等同普通授权恢复旅程。
4. _check_python_forbidden_imports 区分 Import/ImportFrom，后一支丢失导入成员，使同一个子模块依赖得到不同结论。文件有 381 行生产逻辑，同时维护解析、扫描、IO 和策略，不能以 268 行新测试证明整体责任合适。

对主干的风险

独立运行新增测试 13 过，现有 bootstrap/contract 与架构边界测试 601 过、1 跳过；该跳过是同一原生 Windows/PowerShell 用例，base/head 的早期条件、源码和原因相同。原生 premerge 的 6 项直接、6 项 catalog 和 8 项风险检查通过，没有手工 hold；这些绿色结果没有覆盖上述反例。按当前策略没有读取、轮询或等待 CI。

同输入生产路径	旧基线	当前完整 head
普通 bootstrap 后默认 CLI check，无显式工程启用	exit 0，干净	exit 1，软件规则错误
当前内部显式停用软件 probe	旧版无该参数	通过，但 bootstrap 指令/快照仍已产生
同一禁止子模块的直接 import / from-import	尚无此规则	前者拒绝，后者错误通过

两臂都使用真实 bootstrap、公共 CLI 和一次性 File 注册/runtime，未修改活跃 Goal。首次夹具放在被忽略私有树里，两个 CLI 都因独立的 public-boundary 扫描拒绝；原失败保留，随后在全新系统临时目录完整重放两臂，没有删除或关闭生产扫描来制造结果。当前两个反例均稳定成立。修复后应重跑默认/显式开关、完整生成指令和规则元数据、来源变更/删除/合法恢复，以及直接/from/相对导入的正反矩阵。

语义与 CI 对齐

本 PR 新建了来源种类、规则 kind、快照字段及健康诊断语义，kind: str 加注释不是 typed 规则合同；静态 advisory 未报 candidate 也不能证明没有新词汇。软件策略应归显式工程 owner，通用来源完整性应复用当前 typed 边界和诊断语义。include_architecture_invariants 是机器执行分支，失败会产生健康错误，不宜表述成仅供参考的指导。现行原生校验通过仍不能覆盖默认范围和导入漏检。

我的整体评价

REQUEST_CHANGES。来源摘要锁有可验证的价值，但完整提案仍包含默认行为回归和错误的规则通过，用户的正常检查及后续恢复没有形成可用闭环。长期有效工作是否保持尚未证明；用户体验已观察到无额外启用却新增拒绝。来源完整性与软件源代码约束属于两项不同责任；后者不能仅因发现文件名就在通用健康路径默认启用。保留有用的来源校验，先把软件扫描移到显式工程检查边界，并修复已声明导入规则的漏检；无需等待整套控制面迁移。已检查 bootstrap 生成的 Agent Markdown、CLI check 和 status collection：本 PR 没有修改渲染页面组件，问题在默认策略和诊断，因此不以截图充当其正确性证据。未验证范围：原生 Windows/PowerShell、完整跨平台 CI、已安装客户端、长期扫描成本和获授权锁更新后的真实继续工作未获本次资格；这些不替代已复现的两项缺陷。这是当前精确 head 的评审结论，不要求用户代作者操作，也不授予合并、安装或额外执行权限。

English verdict: REQUEST_CHANGES 576af8b. Ordinary bootstrap/check implicitly enables the claimed optional software probe, and a normal from-import bypasses the declared forbidden submodule rule. Independent CLI counterexamples reproduce despite passing focused/regression tests and native premerge; separate the responsibility/activation boundary and repair import resolution.