[platform reading] PR-body write eats a trailing --- horizontal rule AND the attribution footer after it — measured in both repos; issue comments unaffected #12909

Activity

1. os-litant

   added

   pm:queue

   domain:skills

   on Aug 28, 2026
2. os-litant

   added the

   Task

   issue type on Aug 28, 2026
3. os-litant commented on Aug 28, 2026

   

   os-litant

   on Aug 28, 2026 – with Claude

   CollaboratorAuthor

   More actions

   Claim: Director/skills seat, session session_01MnijPVVDakqK2J335JoJtq
   Branch: claude/issue-12909-prbody-sanitizer-fact
   Worktree: ../objectstack-12909
   Domain: skills
   File surface: .claude/skills/pm-dispatch/references/platform-readings.md only — one fact row beside the existing sanitizer entries (314/314 zero-headroom, net 0 + cut ledger).
   Container & model: opus — references/** carries no path mandate (derived live today; mandate globs unchanged through the day's landings). Clause-② limb: no.
   Serial: cleared — platform-readings queue ② (the label-read row) merged 14:10Z; this is queue ③, the last standing entry for the file. Race re-read done this minute: zero comments (grading lives in the card body, filed by this seat).

   ---

   Generated by Claude Code
4. os-litant

   added

   pm:dispatched

   and removed

   pm:queue

   on Aug 28, 2026
5. os-litant

   self-assigned this

   on Aug 28, 2026
6. os-litant commented on Aug 28, 2026

   

   os-litant

   on Aug 28, 2026 – with Claude

   CollaboratorAuthor

   More actions

   os-dev-report

   {
     "issue": 12909,
     "status": "done",
     "branch": "claude/issue-12909-prbody-sanitizer-fact",
     "pr_number": 13007,
     "pr_url": "https://github.com/objectstack-ai/objectstack/pull/13007",
     "pr": "https://github.com/objectstack-ai/objectstack/pull/13007",
     "draft": true,
     "net_lines": 0,
     "premise_still_valid": true,
     "stopped": false,
     "summary": "One fact row landed in .claude/skills/pm-dispatch/references/platform-readings.md as a third sibling in the write-side sanitizer cluster (issue body / comment / PR body), so the naming convention itself carries the cross-reference at zero line cost. The row states: a PR body ending in a --- horizontal rule followed by the attribution footer loses both while the write call reports success; re-writing the footer without the rule survives; issue comments are unaffected in either spelling, so the failure is both spelling-conditional and surface-conditional and a footer verified in a comment proves nothing about a PR body; the operative rule is a bare footer plus a body read-back, which is the only detection. Premise was re-verified rather than inherited: both cited bodies were read on origin/main and both end with a Provenance heading, the session URL as durable prose, and a bare footer with no preceding rule, created one minute apart on 2026-08-28. File is 314/314 with 14 insertions and 14 deletions.",
     "cut_ledger": [
       {
         "cut": "The 写侧 sanitizer 作者规则 row, 6 lines to 3 (saves 3).",
         "why": "It restated the author rule, whose home is AGENTS.md under 'GitHub mutates body BYTES' — fences do not protect, spell poison-shaped tokens in words, read the body back after any less-than fragment. AGENTS.md in the same breath delegates the measured shapes to this table, so the division of labour was already written down and this table held a narrow copy of the other half. .claude/agents/os-dev.md states the same rule explicitly: 一条规则一个家;此处不再复制窄版.",
         "kept_because_this_is_their_only_home": "the entity spelling for a literal angle bracket, and bare identifiers surviving (what keeps the SKILL extraction contract valid)",
         "surviving_homes": [
           "改花括号占位符 / 整句用词描述 -> AGENTS.md ('or are described in words'); os-dev.md carries the placeholder spelling 一律改占位词拼写",
           "⛔ 评论不豁免 -> the next row in this table is the comment-truncation measurement itself, and the new PR-body row now states the surface-conditionality explicitly",
           "⛔ 只回读标记不算验证 / 要回读尾部 -> os-dev.md 干净收尾: 写完读回那条评论到尾部, and PR 正文同欠一次全文回读"
         ]
       },
       {
         "cut": "The credential-discrimination clause in the REST-availability row (saves 1).",
         "why": "references/rest-channel.md already carries it verbatim in its 通道边界 block: repo-scoped probe first, then 403 leads to one /rate_limit call, 15000/hr means live credentials refused by repo-scoping while 60/hr or an auth error means no credentials. That file delegates only 按班矩阵 and 降级梯 back to this table, and both stay here untouched. Replaced by a pointer.",
         "genuinely_lost_and_stated_not_hidden": "the parenthetical that one seat's token measured as a 14-byte placeholder string, and the phrase 两形同症不同治. The operative discrimination they decorate survives in full at the named home."
       }
     ],
     "no_rewrap_line_buying": "Both cuts delete content that is stated in full elsewhere. The re-wrapping visible in the diff is the consequence of removing bytes from the middle of a wrapped paragraph, not a reflow to buy lines. Widest line in the file measures 120 bytes, at the cap and not over it.",
     "footer_survival_measurement": {
       "what_was_written": "PR 13007 body ends with the bare attribution footer and NO preceding --- rule, exactly the spelling the landed row prescribes.",
       "read_back_result": "SURVIVED VERBATIM. Full body read-back immediately after create returns the bare footer intact at the end of the authored text. This is a fourth measurement in the same class and it agrees with the row.",
       "second_finding_same_read": "The platform then APPENDED its own copy beneath the surviving one — two blank lines, a --- rule, and a second footer in the session-URL spelling this seat never wrote. That trailing rule-plus-footer was NOT eaten.",
       "what_it_sharpens": "The eat is not unconditional. What was measured as eaten is the AUTHOR-WRITTEN rule-then-footer present in the payload at create time; the platform's append is added after whatever eats it has already run. So the fact is not 'trailing rules get eaten', and the row's operative rule (bare footer plus read-back) is what actually covers both shapes. A seat reasoning from 'I saw a trailing rule survive' to 'my rule is safe' would be reading the platform's append as evidence about its own write.",
       "practical_corollary": "A bare footer at the tail of a PR body with a rule above it is very likely the PLATFORM's, not yours downgraded — the caution AGENTS.md already states for issue comments now has a PR-body instance.",
       "not_landed_in_the_row": "Deliberately. The file is 314/314 and this nuance needs its own cut. The row as landed does not claim the platform's append is eaten, so nothing here contradicts it and no re-edit is owed. Recorded in full as a comment on PR 13007 and flagged here for the PM.",
       "measurement_artifact_preserved": "Recorded as a PR COMMENT, not a body edit — editing the body would destroy the very artifact the measurement is about."
     },
     "gates": [
       {"name": "check:pm-skill-ratchet (ceiling)", "verdict_line": "✓ check-skill-line-ratchet: .claude/skills/pm-dispatch/references/platform-readings.md is 314 lines (ceiling 314; headroom 0).", "exit": 0},
       {"name": "check:pm-skill-ratchet (max-line)", "verdict_line": "✓ check-skill-line-ratchet self-test: 111 cases pass. — includes ✓ budget is 120 bytes; widest line in the file measures 120", "exit": 0},
       {"name": "check:pm-skill-id-lint", "verdict_line": "✓ check-skill-id-lint: 23 file(s) clean (pattern /#[0-9]{3,}/g).", "exit": 0},
       {"name": "check:pm-governed-prose", "verdict_line": "✓ check-governed-prose: 2 instruction surface(s) name all 5 registered governed surfaces (docs/adr/** · .claude/** · skills/** · AGENTS.md · CLAUDE.md) and claim no others.", "exit": 0},
       {"name": "check:skill-frame-sync", "verdict_line": "✓ check-skill-frame-sync: 4 copies of the decision frame are structurally isomorphic across 3 files", "exit": 0},
       {"name": "check:skill-frame-freshness", "verdict_line": "✓ check-skill-frame-freshness: the decision frame in this tree is current with origin/main (fetched just now).", "exit": 0},
       {"name": "check:pm-governed-merges", "verdict_line": "✓ check-governed-merges --self-test: 206 assertions ...", "exit": 0},
       {"name": "check:agent-test-spelling", "verdict_line": "✓ check-agent-test-spelling: 0 violations — 395 file(s) · 4455 bare -- token(s) · 1214 launcher-rooted run(s) · 9 separator(s) JUDGED", "exit": 0},
       {"name": "check:doc-authoring", "verdict_line": "✓ doc authoring guard: 392 files clean — no bare metadata literals.", "exit": 0},
       {"name": "check:doc-formula-expressions", "verdict_line": "✓ check:doc-formula-expressions: 22 record-scoped formula example(s) across 425 files / 1453 TS blocks judged clean by @objectstack/formula.", "exit": 0},
       {"name": "check:nul-bytes (every-edit convention)", "verdict_line": "check-nul-bytes: OK (scanned 7220 text file(s) -- 7220 tracked, 0 untracked-not-ignored; skipped 7 binary; no raw ASCII control bytes).", "exit": 0},
       {"name": "check-governed-queue-guard (NOT MEASURED, not red)", "verdict_line": "⛔ Governed Surface Queue Guard: could not read GITHUB_EVENT_PATH — its own text says 'could not look' must never exit 0 here; it runs with the event payload in CI.", "exit": 1}
     ],
     "tests": "Union derived mechanically with `node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack` (it read the change set from git itself; 8 path-matched families) and run under `scripts/pm/os-verify-lock.sh` with slot issue-12909, AFTER the final commit, on head 2e9ff3e2b. Exit codes captured by redirecting each gate to its own file BEFORE any pipe — no tail status read as a gate verdict. Verdict lines above are the gates' own printed lines, not a bare $?. The ratchet reading is quoted POSITIVELY: the gate names this file and states 314 lines against a ceiling of 314. `@objectstack/lint` was built inside the same locked run because check:doc-formula-expressions requires it; the quoted verdict is the post-build reading. No changeset owed (one .claude markdown file, releases nothing).",
     "labels": "skip-changeset applied via the MCP fallback because this seat's repo-scoped REST probe returned 403 (GitHub access is not enabled for this session), so the additive POST endpoint is unreachable. Read current values via the zero-quota public-repo payload channel (data-name= spelling, the PR-side one) = {size/s}; union-wrote {size/s, skip-changeset}; comparative read-back returned {documentation, size/s, skip-changeset} — nothing from the union missing, and documentation was added by the labeler after my read, not stripped by my write; delayed second read ~3 minutes later still returns all three. PR confirmed still draft.",
     "mcp_calls": "12 — issue get, issue comments, two PR reads for premise verification, create PR, PR read-back, label write, PR comment, two searches (dedup + positive control), finding create, this report comment.",
     "out_of_scope_findings": [
       "filed as #13008: references/rest-channel.md header quotes a stale default read order (git -> REST -> MCP/GraphQL) while the source it names by name, platform-readings.md, has said git -> payload -> REST -> MCP/GraphQL since 2026-08-25 — the omitted tier is the zero-quota payload channel, and the population that needs it most is exactly the 403-gated seat the header degrades for. Unassigned, unlabelled, left for PM triage; not fixed here (different defect class, and rest-channel.md is a separate governed surface at 87/87 zero headroom with its line 3 already at 118 of 120 bytes)."
     ]
   }

   Generated by Claude Code

   ---

   Generated by Claude Code
7. os-zhuang

   closed this as completedin #13007on Aug 29, 2026
8. os-litant

   removed

   pm:dispatched

   on Aug 29, 2026