Repository navigation
Sharpen CLAUDE.md wake-compliance stanza + fix settings churn (v0.8.1) - #5
Merged
Merged
Conversation
Delivery of DMs is a mechanism (0.7.0) and visibility is automatic (0.8.0), but arming the Monitor and ACTING on a received DM are irreducibly the agent's job — a hook cannot call the Monitor tool. Rewrite the managed stanza to amplify exactly those two steps: arm-first-thing with the precise Monitor invocation, and "a FLEET DM is a task to ACT on, not an FYI" (delivery is guaranteed; arming only changes latency). Point to `fleet.sh monitors` for reachability. Also fix a perpetual churn: `init` emits pretty-printed settings.json while the checked-in file was compact one-line arrays, so every `init` re-dirtied it with a whitespace-only diff. Commit the installer-canonical pretty form → `init` is now byte-idempotent on settings.json (verified by two consecutive runs). Verified: shellcheck --severity=warning clean (CI's exact invocation); selftest-collateral OK (every control bites, incl. SETTINGS + LIVENESS); `fleet.sh init` refreshes the stanza in place. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The agent-facing capstone of the wake-reliability work.
Delivery of DMs is now a mechanism (v0.7.0) and visibility is automatic (v0.8.0). But two steps are irreducibly the agent's job — a hook physically cannot call the Monitor tool: arming the watcher, and acting on a DM once received. Compliance is the only lever left for those, so the managed
CLAUDE.mdstanza is where to amplify it.Changed
persistent: true,timeout_ms: 3600000), and a note that the⚠ FLEET WAKE NOT ARMEDre-nudge (v0.8.0) fires every turn until done.fleet.sh monitorsfor reachability before relying on a wake.Fixed
initno longer re-dirties.claude/settings.jsonevery run. The checked-in file was compact one-line arrays whileinitemits pretty-printed JSON, so each run produced a whitespace-only diff (drift that masks real changes). Committed the installer-canonical pretty form →initis now byte-idempotent on settings.json (verified: two consecutive runs = identical output).Verified
shellcheck --severity=warning install.sh .fleet/bin/*.sh→ exit 0 (CI's exact invocation).selftest-collateral.sh→ every control bites, incl. SETTINGS (hooks still wired after reformat) + LIVENESS.fleet.sh initrefreshes the marker-guarded stanza in place; the checked-inCLAUDE.mdmatches its output.🤖 Generated with Claude Code