[Weimr2627]

Goal And Delivered Outcome

• Outcome basis / optional anchor: reproduced defect [Bug]: explore-todo-result-node-linkage-smoke fails on main — the archive section drops a token the readback still returns #8134, plus the accepted projection change PR feat(control-plane): sqlite_v0 bounded active view projection, in-transaction auto-archive, and goal-scoped stale_source_registry #7834 (feat(control-plane): sqlite_v0 bounded active view projection, in-transaction auto-archive…, merged 2026-11-02T18:20:05Z).
• Goal/source and gap: examples/explore-todo-result-node-linkage-smoke.py fails deterministically on main. #7834 capped the inline sqlite_v0 archive rows and replaced them with a pointer that names the readback (query="loopx todo list --goal-id <goal-id> --include-archived"), and it changed loopx/control_plane/todos/machine_section_projection.py by +377/−5 without touching this shipped smoke. The smoke still demanded the pre-feat(control-plane): sqlite_v0 bounded active view projection, in-transaction auto-archive, and goal-scoped stale_source_registry #7834 inline repetition, so one consumer was left behind while the guarantee it protects still holds through the readback.
• Observable before → after, with the validation row that proves it: before, the sqlite arm raises at the inline-token assertion; after, both file and sqlite arms pass and the sqlite arm asserts the pointer (archived_count=1 and the goal-scoped --include-archived query) instead. Proven by the smoke row and the mutation rows below.
• Issue/task and intended base: Related to [Bug]: explore-todo-result-node-linkage-smoke fails on main — the archive section drops a token the readback still returns #8134 (left open deliberately — which surface should render archived rows is the maintainers' call, and this PR changes only the consumer). Intended base is loopx-project/loopx:main.

Onset evidence, not guesswork

Reading	Result
nightly run 41356265689 (2026-11-01T22:16:06Z)	smoke absent from the failed log (log retrieved: 1,097,104 bytes / 9,555 lines, 0 mentions) → green
nightly run 41557935873 (2026-11-03T01:10:15Z) and 41740364791	smoke present 11 times in each failed log → red
commits touching the projection in that window	all from #7834; this smoke got no commit in the window
local reruns on 4d8b79692d13ea17a43d2369d35282b023be311e	failed twice at the same assertion

Author Declaration

• Written by: model_agent — Qoder AI assistant, at the contributor's direction; no human review has occurred.

Implemented against

• Specification and revision: loopx/control_plane/todos/machine_section_projection.py format_sqlite_archive_pointer_comment() at 4d8b79692d13ea17a43d2369d35282b023be311e, as introduced by merged #7834; the linkage guarantee recorded in Explore planner: add explicit todo-to-result-node linkage before typed evidence scoring #1819.
• Criteria:

Criterion (spec clause)	Disposition	Symbol / path	Test or command
Archived linkage survives todo archive-completed under file authority	implemented	examples/explore-todo-result-node-linkage-smoke.py:189	the smoke
Archived linkage survives under sqlite_v0 through the pointer the projection emits	implemented	same file, :172-187	the smoke
The pointer names the readback that returns the record	implemented	:185	the smoke
The archived record still carries explore_result_node_refs	implemented (unchanged)	:193-194	the smoke
Projection production code changed	out_of_scope	loopx/control_plane/todos/machine_section_projection.py	N/A — consumer-side reconciliation only

• Self-check before submission: read the pointer builder, confirmed #7834 is the only in-window change to the projection, confirmed #8013's diff on that file is typing/cast only (+4/−3) so it is not already repairing this, and left the assertion order untouched elsewhere. Each new guard was poisoned once and every poison was caught (rows below); the file was restored byte-identical after each run. No production behavior, fixture data or provider semantics were changed.

Scope And Continuation

• Completed scope and remaining work: this closes the consumer drift for this one smoke. The unresolved question in [Bug]: explore-todo-result-node-linkage-smoke fails on main — the archive section drops a token the readback still returns #8134 — whether sqlite_v0 should still render archived rows inline — is a projection decision and stays with maintainers; if they rule the other way, this smoke change should be reverted rather than the projection kept.
• Slice boundary / successor: single shipped smoke, independently reversible, no successor needed for the red nightly lane.

Validation

• Tested revision: 24ccac82d00ced2bedb53e72236f2b2aac091713
• Run state: finished
• Input classes: synthetic

Check kind	Result	Public-safe evidence / limitation
integration	passed	PYTHONUTF8=1 python examples/explore-todo-result-node-linkage-smoke.py → explore-todo-result-node-linkage-smoke ok, both file and sqlite arms.
regression_parity	passed	On base 4d8b79692d1 the same command fails at the inline-token assertion (two local reruns, same line) and nightly records show green 2026-11-01, red from 2026-11-03; the file arm passes on base, which is the positive control that the drop is sqlite_v0-specific.
regression_parity	passed	Mutation control, one poison per new guard, each caught at its own line: archived_count=1→7 (line 183), len(pointer)==1→5 (182), the readback query token (185), and the file arm inline token (189). No poison survived, so no guard here is vacuous.
static	passed	python -m py_compile; git diff --check.
unit	not_run	No unit test covers a shipped smoke's prose assertions; the smoke is its own harness.

• Coverage and gaps: the diff is a test-only reconciliation, so the covered surface is exactly this smoke's lifecycle path (add → complete → archive-completed → readback). Not covered: the rest of the nightly fleet, and the #8134 question of whether the projection should cap those rows. This workflow is intentionally not PR-required, so no hosted run is attached to this head.

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: synthetic
• Attention review: N/A
• Craft review: N/A

Type of Change

• Bug fix
• Test update

LoopX Area

• Control plane (goals, todos, quota, scheduler, registry, runtime)

Technical Direction

• Direction / acceptance reference, when applicable: Core control-plane hardening; the sqlite_v0 bounded projection introduced by feat(control-plane): sqlite_v0 bounded active view projection, in-transaction auto-archive, and goal-scoped stale_source_registry #7834.

Shared-authority RFC fixture impact

• Production-scale fixture schema: N/A
• Semantic dimensions changed, or reviewed no-impact rationale: N/A — no fixture dimension, provider arm or schema changed; one shipped smoke's archive assertion was aligned to the projection already on main.
• Provider conformance arms run: N/A — the smoke's existing file and sqlite arms both run; neither was added or removed.
• Read-only legacy/file/PostgreSQL three-arm rehearsal (required for promotion, runtime-routing, or compatibility-projection changes): N/A — no compatibility projection is introduced by this diff.

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). Fixture digests and temporary paths are omitted from the evidence above.
• 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 RFC changed here).
• 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: actor_kind=model_agent; model=gpt-6.1-sol; provider=OpenAI; declaration_source=runtime_reported; reasoning_effort=max.

动机 / Motivation

维护 Explore 证据关联的贡献者会遇到一个重复失败的公共验证。

SQLite 已采用有界归档投影，正文留下查询指针；旧 smoke 却仍要求内联重复证据引用，导致真实数据仍可读时验证提前失败。

这次修复使既有 File 与 SQLite 两条归档验证都能完成，同时继续核实归档后的原始证据引用。

边界是修复一个现有验证消费者，不改生产投影、不恢复大块内联归档，也不证明整套 nightly 或长期收益。

改动思路 / Approach

没有阻塞问题，另有一项可直接收紧的 P3 断言建议。先读了原始 #1819 的关联保持契约、#8134 的问题与补充，以及不可变基线上的投影 owner，再判断补丁。源码明确把 SQLite 归档显示与 canonical record 分开：有界显示只留指针，完整记录仍由正常 Todo 读回提供；File 保持原内联断言。因此，仅调整这个既有 smoke 比扩大生产投影或删除归档检查更合适。

独立执行了指针中给出的 todo list --goal-id … --include-archived 查询，真实 SQLite 返回一个归档记录和原来的两项 Explore 引用。这里的断言不是用 pointer 存在代替数据存续，也没有把显示文件变成权威。现有 provider 投影、完整归档读取和 durable-link 测试分别保护相邻边界，未发现需要新增另一个 smoke 的理由。

具体改动 / Changes

准确 head 为 24ccac82d00ced2bedb53e72236f2b2aac091713，完整 diff 仅 examples/explore-todo-result-node-linkage-smoke.py +18/-1；没有生产或 fixture 改动。

spec_ref 为 https://github.com/loopx-project/loopx/issues/1819，spec_revision 为 a4370d644af3932b8c6aef62f8d161aed709dfc9，表示本次判断所核对的既有实现基线。原 issue 无条款编号，以下仅作映射：LINK-PRESERVE 保留 add、complete、archive、update/clear 后的显式引用；LINK-EXPLICIT 保留未知/反驳节点诊断和无关联任务原行为；LINK-NO-EFFECT 保留 diagnostic-only、零分数调整及不启动 Agent、不取得 claim/lease、不改 quota。这三项均由完整 smoke 的真实 CLI 生命周期和末尾 planner 断言通过。

关键内容讲解

SQLite 分支在 check_todo_lifecycle 中找到带当前 Goal 的唯一指针，核对计数和完整查询文本；File 分支继续要求内联 token。后面的 canonical Todo 读回仍要求 archive 状态和完整原引用，原有 clear 与 planner 诊断也继续执行。对已有 smoke 的这一个消费者做修复，保留了可单独回退和复现的边界。

对主干的风险 / Risks

P3：计数应匹配完整 token。 第 183 行用 "archived_count=1" in pointer[0]，也会接受 archived_count=10。我在真实 CLI 完成归档后，只改动其临时显示文件的计数，整条 SQLite smoke 仍通过；改成 7 则被拒绝。建议匹配带字段边界的 token 或解析属性，并加入 10 的反例。这是新增的次要计数保护不够精确，不是已发现的 canonical 数据损坏；实际指针查询和引用保持均独立通过，故不作为本次合并阻塞。

PASS：准确基线的完整 smoke 在旧 SQLite 内联断言失败，当前 head 的 File/SQLite 全流程通过；独立跟随真实指针查询成功。临时投影中的缺失当前 Goal 指针、错误查询、File 引用缺失和计数 7 四个反例均被相应断言拒绝，未修改生产或活动 Goal。编译、diff hygiene、scoped CQR 与 quick premerge 通过。

原始 maintainability 检查仍 FAIL，完整 base/head JSON 仅去掉 checkout 路径后相同，保留既有 62/60 decision-points 失败；没有放宽阈值。没有查询、轮询或等待 CI。作者的历史 nightly 数字没有作为本次独立验收证据。

我的整体评价 / Verdict

goal_achieved 仅指这个可复现的 smoke 消费者漂移；long_horizon 为 preserved，现有证据关联、诊断与权限边界保留；user_experience 为 improved，贡献者可以完成原有验证，无新增配置或人工交互。P3 计数前缀漏检明确保留，推荐顺手收紧，未把它写成完整的计数敏感性通过。

这是对已有 322 行 smoke 的局部更新，没有新建临时 builder 或重复生命周期测试。视觉检查不适用：没有页面、布局或产品显示改动；未验证范围是其余 nightly、线上宿主采用和完整长期成本。批准与合并、安装、发布权限分开。

English verdict: APPROVE — 24ccac8 repairs a reproduced smoke-consumer mismatch while preserving File/SQLite linkage and diagnostics. Real pointer readback and negative controls pass. Tighten the nonblocking P3 count-prefix assertion; the independently matched ratchet failure remains failed.

[loopx-agent]

[P3] 请让归档计数匹配完整字段："archived_count=1" 也会命中 archived_count=10。在真实 CLI 完成归档后，只将其临时投影中的计数从 1 改为 10，整条 SQLite smoke 仍通过；改为 7 则被此行拒绝。建议解析属性或匹配字段边界，并补充 10 的反例。实际 canonical 引用与指针查询均已独立通过，此项是非阻塞的测试敏感性建议。