docs(handoff): cairn slice 1 MERGED and switched, slice 2 open as #1406 with its round-1 🔴 fixed but uncommitted

[ZacxDev]

Updates claudedocs/handoff-cairn-oss-multi-instance.md for the rank 3 half-2 work.

What moved since the last revision (b3c9a93d):

• devrc refactor(cairn): who becomes its own cairn-who binary, and the shared timeout predicate moves to the module both import #1381 MERGED (baa664e4) — cairn who re-homed as cairn-who, unbounded_timeout_reason extracted to scripts/lib/timeouts.py. Three audit rounds; the ladder ended on the attribution gate (two consecutive zero-payload fix rounds), not a clean round. ship.sh converged BOTH hosts; verified live.
• devrc docs(handoff): the cairn lib/ fork is DECIDED — consolidate onto the pin #1394 MERGED (65d8bfba) — records the operator decision CONSOLIDATE ONTO THE PIN. The fork had said "unanswered" in three places; all three fixed, because one stale copy re-opens a settled question.
• Slice 2 is PR feat(nix): deploy cairn from the PINNED flake package, not a fork in this checkout #1406, pushed. Round-1 audit found one 🔴, one 🟡, six 🟢; all fixed but UNCOMMITTED in ~/workspace/devrc-flake-pin — round 2 not yet run.

Three new ranked items (16–18): the gate never executes the deployed binary; the stale who comment still live in the public repo; three deferred #1406 audit findings.

Carry-forward note. State now and Next steps are REPLACE sections and the merge warned that 10 durable lines would be dropped. I rewrote State now to carry them forward — cairn's six merged PRs, rank 12's closure, ranks 4/8/13/15, rank 11 still refused, the unpublished image, session capture, the opencode exporter — and re-ran until the remaining 8 were all restatements my new text carries in different words. Nothing was deleted on the strength of "it's stale".

Landed on a branch, not main. The base clone sits on main and devrc is not a trunk-deploy repo, so the doc was merged in a dedicated worktree (docs/handoff-cairn-slice2) and pushed there.

🤖 Generated with Claude Code