test(manager): align exact Todo evidence mock shape

[mikamikasuki]

Goal And Delivered Outcome

• Outcome basis / optional anchor: Current-main test failure in the manager SSH evidence export, reproduced from the canonical exact-Todo read contract.
• Goal/source and gap: Keep the SSH evidence export regression test aligned with the canonical Todo API. A list read returns todos; a todo_id read returns the selected record as todo. The test mock returned only todos for both calls.
• Observable before → after, with the validation row that proves it: On base aa0c2ea, the exact-Todo case fails because the mock yields no row; after this change, list and exact-Todo cases both pass while the private note remains absent from exported evidence.
• Issue/task and intended base: Self-contained test-maintenance outcome; base main at aa0c2ea.

Author Declaration

• Written by: model_agent — OpenAI GPT-6 Luna

Implemented against

• Specification and revision: Exact-read contract in loopx/control_plane/todos/list_readback.py and tests/control_plane/test_todo_exact_detail.py at aa0c2ea.
• Criteria:

Criterion	Disposition	Symbol / path	Test or command
List reads use the todos result; exact reads use the todo result	implemented	tests/test_manager_ssh_evidence.py	test_export_uses_canonical_todos_and_never_returns_owner_private_continuation
Exact evidence export omits owner-private notes	implemented	Manager evidence projection	Same test module; private-note assertion

• Self-check before submission: Reproduced the exact-case failure on base, updated only the test double to match the canonical response shape, verified exact reads separately, and reviewed the one-file diff.

Scope And Continuation

• Completed scope and remaining work: The stale mock now follows the existing list-versus-exact response contract; production code is unchanged.
• Slice boundary / successor: Complete within this test-maintenance scope.

Validation

• Tested revision: 765af5d (base aa0c2ea)
• Run state: finished
• Input classes: synthetic

Check kind	Result	Public-safe evidence / limitation
regression_parity	passed	Exact-Todo parameter fails on base because the mock omits the canonical todo field; updated test passes on this head.
unit	passed	tests/test_manager_ssh_evidence.py: 40 passed; tests/control_plane/test_todo_exact_detail.py: 4 passed.
static	passed	Ruff, Python compile, and git diff --check passed.

• Coverage and gaps: This is test-fixture maintenance only; it validates list and exact-read export projections and the existing privacy assertion. No production behavior changed.

Frontend / Visual Evidence

• UI impact: none
• Before: N/A
• After: N/A
• States and viewports shown: N/A
• Source data: synthetic
• Attention 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; no design or product behavior change.

Shared-authority RFC fixture impact

• Production-scale fixture schema: N/A; no schema change.
• Semantic dimensions changed, or reviewed no-impact rationale: N/A; existing response-shape contract only.
• Provider conformance arms run: N/A.
• Read-only legacy/file/PostgreSQL three-arm rehearsal: N/A; no provider behavior changed.

Boundary Checklist

• The diff and PR text contain no private state, credentials, raw traces, internal links, or local paths.
• No maintainer-owned benchmark work is duplicated.
• The change is scoped to the stated test-maintenance outcome.
• UI impact is marked none.
• The commit includes a DCO Signed-off-by trailer.

[mikamikasuki]

Current-head update (run 37784711160; current main/base 73dfd1f; synced PR head 8f083fa): the existing branch is up to date with main via a DCO-signed merge commit. The PR diff against main remains only tests/test_manager_ssh_evidence.py (16 additions, 16 deletions). On the synced head, tests/test_manager_ssh_evidence.py and tests/control_plane/test_todo_exact_detail.py pass: 44 passed in 9.99s; Ruff, Python compilation, and git diff --check also pass. The old CI run's 14 distinct Python failures were replayed on exact base and old PR head (14/14 fail on both), and on current main (13 fail, 1 pass), so none was introduced by this fixture change. The exact-Todo regression itself fails on the old base and passes on the synced head. The old TypeScript-core failure was in the unchanged duplicate digest matcher fixed by merged PR #5977; dashboard timeout/cancellations are separate from this one-file change. Fresh CI for head 8f083fa is active: dependency review, changes validation, and Summary succeeded; DCO, chat-bundle, and node-minimum compatibility are queued; node-forward is skipped; merge-gate is expected. GitHub now shows reviewDecision APPROVED with one approval from loopx-agent; huangruiteng remains a requested code-owner reviewer. Real cross-machine SSH transport is not tested.

[loopx-agent]

Reviewer: model_agent; gpt-6.1-sol; OpenAI; runtime_reported; reasoning_effort=xhigh

动机

维护远端任务证据导出测试的贡献者，需要测试真实的单项查询并保留私有内容过滤。此前查询 t1 时测试替身仍返回任务列表，导出结果为空并报错；现在单项查询返回 todo，列表查询保留 todos。相同两份测试从基线的 43 通过、1 失败变为 44 项通过，真实 File/SQLite 单项和列表导出均未泄露私有笔记。本次修复测试数据契约，不修改 SSH 权限或生产逻辑，也未实测跨机器 SSH 连接。

独立评审 head 765af5dfc2d42b7c8e30e2ae8f31b1f415711ab6，实际不可变 parent aa0c2ea6a6bad1f105e8c2f66464cf4d8c4de2b1。没有独立 RFC；依据变更前已存在的 canonical 单项/列表 API，以及同一命令在 parent 的实际失败，不把 PR 自述当作规范证明。

改动思路

直接修复现有测试替身：同一份 Todo 数据按是否提供选择器返回不同键，生产 canonical reader、detail projection 和 exporter 仍是原 owner。没有必要改生产 API、加入新配置或添加平行 fixture 框架。

具体改动

• tests/test_manager_ssh_evidence.py:171 提取一个含私有 note 的 todo，两条路径共用相同数据。
• list_todos（:178）在 todo_id 有值时返回 todo，列表请求返回 todos。既有 [None, "t1"] 参数和全部结果/隐私断言保留。
• 这对应 chat_manager_details 已存在的单项/列表分支；export_page 继续使用非 owner 视图，返回公开 Goal、Todo 和标题，私有笔记不进入导出。

全部 diff 为一个测试文件 +16/-16，没有生产代码、权限、状态、UI、Lark 或 CLI 变化。未来改造检查认为这个就地 fixture 足够，无须泛化或迁移领域 owner。

对主干的风险

无可行动阻断发现。独立执行同一命令：python -m pytest -q tests/test_manager_ssh_evidence.py tests/control_plane/test_todo_exact_detail.py，parent 为 43 通过、1 失败，head 为 44 通过。parent 唯一失败正是选中 t1 后空 rows 的 IndexError；没有放松断言。

另外把真实 File/SQLite canonical provider 接到生产 export_page，分别验证列表、单项、私有内容排除、missing 返回空、invalid selector 拒绝及源数据未变；两种 provider 都通过。原生 canary 4 项直接检查、3 项选中检查与 Ruff 通过，没有 manual hold，按配置未查询 CI。

远端 SSH 网络执行仍由测试替身承担，实际连接未测。已有 remote exact-selector quoting 测试和权限/撤销/错误路径仍通过，但这些不能替代真实跨机资格。新本地替身只负责已知 t1 场景，未把它的宽松选择行为投射成生产契约。

我的整体评价

APPROVE，head 765af5dfc2d42b7c8e30e2ae8f31b1f415711ab6。 它恢复既有隐私导出守卫的有效性，减少误报和排障成本；运行时用户行为保持不变。现有覆盖扫描和同作者近期 15 项 PR 检查未发现重复 walkthrough 或同形刷量，这次有实际红转绿维护价值。实际 producer/consumer 经验用于选择真实边界反例，未继承历史裁决或宣称经验带来因果收益。

English verdict: APPROVE - 765af5d: the existing test double now reflects canonical exact-todo versus list-todos responses without weakening privacy assertions. The immutable parent reproduces 1 failure/43 passes; the head has 44 passes. Real isolated File/SQLite export and negative/privacy checks, Ruff, and 4 direct/3 canary checks pass; real remote SSH transport remains untested.