Name the wait in the Stop hook's suggested calls, route adopt's counter reset through fsys, and stop writing empty counter files #421

Merged
mesh-tools merged 2 commits from fix/1749-1757-wait-list into main 2026-10-21 18:19:19 +00:00
2 Commits
Author SHA1 Message Date
jochen e3a714d5f0 Stop writing an empty refusal-counter file on every stop of every session (issue 1749)
mesh/merge-gate head e3a714d5f0f0 merged into trunk c3564c1accf3 — pass: builds claude-code → ace, g14, novox, shanks; no bus step; every machine composes …
mesh/repo-check head e3a714d5f0f0 merged into trunk c3564c1accf3 — pass: its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-…
mesh/delivery delivered
writeRefusalCounters wrote "{}" and, since it now routes through fsys
(this branch's earlier commit), created the waits folder on every single
stop, even for a session that never had a wait: one counter file per
session id, piling up on a node that uses the wait list at all. It now
removes the file instead of writing empty counters, a no-op when there
was never one to begin with, so a stop with nothing to record leaves
neither the file nor the folder; a refusal still writes it, and the
stop right after the wait is done (or let through) removes it again.
2026-10-21 19:41:05 +02:00
jochen 08cec8cb1c Route adopt's refusal-counter reset through fsys, and name the wait in the Stop hook's suggestions (issues 1749, 1757)
issue 1749: AdoptName reset both session ids' refusal-counter files with
os.WriteFile, bypassing fsys; on a node whose agent has its own account
(ADR 0266) the module runs as the operator's account and cannot write
into the agent's home directly, so the reset silently failed. writeRefusalCounters
now goes through fsys (MkdirAll then WriteFile), as every other write into
the agent's home does; it is also called from the Stop hook itself, which
already runs as the agent's own account, where fsys is its zero value and
behaves exactly as the plain write did before.

issue 1757: the Stop hook's suggested watched and said calls left out `by`,
which claude_code_wait requires on every op, so a session following the
suggestion word for word was refused. Both now carry "by" from the wait's
own By field (already kept per wait since issue 1629). The SessionStart
hook's suggested adopt call already carried by; added a test for it too,
alongside the Stop hook's two, each extracting the exact JSON the hook
prints and feeding it back to the tool's own dispatch to prove it is
accepted as printed.
2026-10-21 19:35:33 +02:00