platform-readings.md third increment: the two post-ruling channel facts (+11), #13387's CDN-cache cause, and #13257 — needs its own ceiling ruling before dispatch

[claude]

Filed by the domain:skills seat (session session_019RfFHiRCSs3JXLK4cwcfox, os-steve) from the review of PR #15190 (#14133 second increment), which landed the eight ruled readings at exactly +34 (ceiling 358, headroom 0) and measured what it could not land.

What is owed, measured

• The two post-ruling facts recorded on [finding] The &page=N completeness walk refuses with HTTP 422 past 9,900 items — the ceiling on the remedy #14068 just landed #14133 after the [PM decision] skills lane batch 7 — raise the platform-readings.md line ceiling by the measured +34 so the intake family's eight readings can land #15013 ruling: the channel-scoped attribution-footer rows (+5) and the separate MCP/REST write-channel rate budgets (+6) — +11 lines in the file's voice (text and placement in PR docs(pm-skill): land the intake family's eight platform readings under the +34 ceiling ruling #15190's body).
• [finding] The zero-quota issue-page payload is CDN-CACHED and can be minutes-to-hours stale — cache-busting and no-cache did NOT defeat it, and one seat asserted a half-state about another seat's work on that stale read #13387's residue: the file states the rule (absence on the payload channel is not a reading; claim-bearing fields never come from it) but not the CAUSE — the channel is CDN-cached, and a cache-buster query string plus a no-cache header were measured across three fetches over 75 minutes and did not defeat it. One line.
• Record the agent-created-Routine transport facts in references/platform-readings.md — fired sessions get no GitHub MCP tools, and self-binding trades new-session recoverability for tool availability #13257: a different fact family, never inside the [PM decision] skills lane batch 7 — raise the platform-readings.md line ceiling by the measured +34 so the intake family's eight readings can land #15013 sizing; entirely unaddressed by the two increments (its facts to be re-measured at dispatch).

How it lands

A ceiling raise is a maintainer-floor act: this card is dispatched only on a ruling that sizes it (a decision card in the skills lane's next batch, recommendation: raise by the measured need as #15013 did). Governed (.claude/skills/**) ⇒ draft PR, both approvers, human merge; the ceiling row cites the ruling; the ordinary-raise-versus-move accounting follows #15191's outcome.

Refs

#14133 · #15013 · PR #15190 · #13387 · #13257.

---

Generated by Claude Code

[claude]

Second measured observation — PR-body footer under a raw REST PATCH /pulls (skills seat, 2026-09-04T06:55Z; for the third platform-readings increment ruled here).

Surface: PR #15073, body edited by the #15061 dev (branch claude/issue-15061-catalog-gates-scan-references) with a raw REST PATCH and a full read-back, 22,533 characters compared.

• The session-URL footer block (_Generated by [Claude Code](https://claude.ai/code/session_…)_) survived the edit verbatim.
• A bare footer block (_Generated by [Claude Code](https://claude.ai/code)_) was appended under it, leaving two blocks; everything else came back byte-identical.

This matches the first observation already recorded on #14133 for the same channel (REST PATCH appends a bare footer, +58 bytes, and leaves a session-URL footer intact; the MCP update_pull_request PATCH is the channel that deletes the block). It diverges from the AGENTS.md clause that says every PATCH normalises a session-URL footer down to bare — that clause describes the MCP channel, not REST. Two independent observations on the REST channel now agree; the AGENTS.md wording is a candidate correction for the increment this card rules on, not an edit from here (governed surface).

---

Generated by Claude Code

[claude]

Claim: PM loop round 5 — measurement-first flight (landing conditional on a ceiling ruling)
Session: session_019RfFHiRCSs3JXLK4cwcfox
Branch: claude/issue-15192-platform-readings-third-increment
Worktree: objectstack-issue-15192
Domain: domain:skills
File surface: .claude/skills/pm-dispatch/references/platform-readings.md only — write the third increment in the file's voice (the two post-ruling channel facts as measured in PR #15190's body, #13387's CDN-cache cause, #13257's three Routine transport facts re-measured from the card, and the second footer observation recorded on this card), wrap with the gate's own wrapLine at 120 bytes, measure the exact line cost per member; open a governed draft PR that carries the lines and STAYS RED on the ratchet (ceiling 358 untouched) until the maintainer sizes the raise. ⛔ No edit to scripts/pm/check-skill-line-ratchet.mjs (the ceiling row and the ruled-raise accounting are a patch round after the ruling and after #15191 lands), ⛔ no other file, ⛔ no deletions from the file to pay for lines (nothing left to delete — measured under #15013).
Container & model: S, mode:subagent, model: opus (tier: no path-derived mandate; prose in the file's Chinese voice under a byte-budget gate)
Clause-②: no (reference prose; no accept/reject behaviour)
Serial constraints cleared: PR #15190 (#14133 family) MERGED 96b4af88 — the file is at 358/358 on main; PR #15057 MERGED 1bc3c092; nothing in flight edits the file (#15191 edits the ratchet script, which this flight does not touch). Governed (.claude/**) ⇒ draft PR, both approvers, human merge; the ceiling raise is a maintainer-floor act — this flight measures and holds, it does not land.

Decision re-read (07:27Z): the card's own text — "dispatched only on a ruling that sizes it" — is honoured by splitting the flight: measurement now (⛔ no fix before the measurement exists), the decision card filed from the measured numbers, the ceiling row raised in a patch round on the ruling.

---

Generated by Claude Code