docs(pm-skill): reconcile platform-readings.md lines 26–28 (merge-tree "does not apply the driver ⇒ under-reports") with the measured 349–350 (driver applied when registered; the hazard is a deferred content judgement)

[os-zhuang]

Filed by the director seat at the #16379 ruling (decision batch #63, 2026-09-07, option A: land the new lines, reconcile the old ones in their own diff).

The contradiction

.claude/skills/pm-dispatch/references/platform-readings.md:

• Lines 26–28 (older, ruled): git merge-tree answers the default text merge and does not apply the .gitattributes merge=os-regen driver, and therefore under-reports conflicts on driver-managed paths.
• Lines 349–350 (PR docs(pm-skill): the fifth readings increment — 26 measured platform facts #16379, measured on git 2.43.0): the driver is applied when the clone has it registered; the hazard is that merge-tree reports clean because the driver deferred the content judgement — it says nothing about content. Registration is per clone, so both observations are true; the inference on line 27 is not.

PR #15885's AGENTS.md corollary (ruled the same day) carries the same corrected reading: probe from a driver-less bare clone, never with -c merge.os-regen.driver=.

Scope

• Rewrite line 27 (and 26/28 if they restate the inference) to the registered-driver hazard, in the register the ledger uses; keep the line count neutral (the ceiling is [Decision] platform-readings.md ceiling 359/362 → 385/388 for the fifth readings increment (PR #16379, red on the ratchet by design) — and whether a standing rule replaces per-increment ceiling cards for this ledger #16386's business, not this card's).
• Cross-check the wording against AGENTS.md §11's corollary once PR docs(agents): a local merge-tree of an os-regen path is not GitHub mergeability — state the corollary and name the sound probe (#15871) #15885 lands so the two governed texts say one thing.
• Governed surface: draft PR, human merge, skip-changeset.

Acceptance

• 26–28 and 349–350 read as one rule with two clauses (registered clone vs driver-less clone)
• no line-count change
• wording agrees with AGENTS.md §11 corollary

Refs PR #16379 (open question), PR #15885, #15871.

[claude]

Blocked-by: #15977

Same file as PR #16379 (the fifth readings increment, approved and armed by the approver, waiting only on its ratchet patch round — flight P2). The reconciliation of lines 26–28 is its own diff by the ruling, so it dispatches the moment that PR merges. pm:queue → pm:blocked for the wait; #16387 closed as this card's duplicate. Skills lane seat (session session_019RfFHiRCSs3JXLK4cwcfox, os-steve), 2026-09-07T02:3xZ.

---

Generated by Claude Code