Everything I could decide myself is decided. What's left needs
you. Each one has a recommendation from Fable — if you agree, pick it and
press Done. "I'm not sure" is a real answer and becomes work for me.
You don't need to open the session — pressing Done sends your answer back and it picks up.
1Merge propflow-docs#6? It fixes a bug where one session running 'docs rm' leaves the shared docs checkout in a state that blocks publishing for EVERY session — it has fired twice in 48h and once took the decisions dock offline. The fix is verified and mergeable. It is not live until you merge it, and the bug is reproducible in the shared checkout right now.
In plain terms. Right now, when any session deletes a page from the docs site, it jams publishing for every other session until a person manually cleans it up. It has happened twice in two days, and once it knocked the decisions page itself offline. The fix is written and tested and just needs your merge: merge it and the problem stops; leave it and the next deletion jams everyone again.
Fable recommends: Merge #6 now — the fix is verified and the defect stays live until it lands The fix leaves the safety guard's behaviour unchanged (the only edit swaps in a shared list holding the identical paths), and both controls passed: the positive one is proven in real git history on the shared checkout, and the operator independently re-ran the negative one — with someone else's unsaved work present, the guard still refused and nothing was deleted or committed. The known cost is that 'docs rm' now writes a commit and a push, which the PR states out loud and the README documents. Every hour it waits, any session running 'docs rm' can jam publishing for everyone. Tried first, unsuccessfully: receipt f5c1b5445 — Fable RESOLVED (held under option-reversal, high/high confidence) that this should be surfaced now rather than left in a passive awaiting-human queue: it confirmed the bug is still live in the shared checkout (bare rm at bin/docs:399, no GUARDED_PATHS), that the repo runs zero CI on PRs so no machine event will ever advance this, and that a recurrence could take down the decision surface itself. It could not settle the merge itself — merging is a rung reserved for a human, not a question an agent may close.