Repository navigation
Layer 4: monitor visibility + arm re-nudge (v0.8.0) - #4
Merged
Merged
Conversation
Surfaces the silent "live-but-unmonitored agent misses wakes" failure and self-heals arming — the operator-facing half of the wake reliability work. - fleet.sh monitors: classifies every agent MONITORED / UNMONITORED (live but no armed watcher) / CLOSED with unread counts, ranked unmonitored-with-unread first + a "will miss wakes" flag. Makes the invisible failure actionable. - Watcher liveness breadcrumb: wake-cmd's watcher now touches $STATE_DIR/wake/<sid>.monitor each tick → robust pgrep-free liveness that also catches a wedged watcher. lib.sh: monitor_fresh + agent_liveness_state + unread_count (fail-safe: MONITORED requires a POSITIVE fresh breadcrumb). - awareness.sh re-nudges to arm the watcher when it's not running (every turn until armed → a lapse self-heals). Detection = mechanism; arming = compliance (a hook can't call the Monitor tool — the irreducible platform limit). Verified: selftest LIVENESS controls (MONITORED/UNMONITORED/CLOSED transitions) pass; fleet.sh monitors classifies + ranks correctly; shellcheck --severity=warning clean. 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 operator-facing half of the wake-reliability work: surface the silent "live-but-unmonitored agent misses wakes" failure, and self-heal arming.
Added
fleet.sh monitors— classifies every agent MONITORED / UNMONITORED (live but no armed watcher → won't proactively wake) / CLOSED, with unread counts, ranked unmonitored-with-unread first + a "⚠ will miss wakes" flag. Turns the invisible failure into a signal the operator can act on (directly answers "how do I know which agents aren't monitoring").wake-cmd's watcher now touches$STATE_DIR/wake/<sid>.monitoreach tick → robust, pgrep-free liveness that also catches a wedged watcher. New lib.sh:monitor_fresh,agent_liveness_state,unread_count(fail-safe: MONITORED requires a positive fresh breadcrumb).Changed
awareness.shre-nudges to arm when the watcher isn't running (⚠ FLEET WAKE NOT ARMED), every turn until armed → a lapse self-heals. Detection is a mechanism; arming stays compliance (a hook can't call the Monitor tool — the irreducible platform limit, stated honestly).Verified
selftest-collateral.shLIVENESS controls (MONITORED / UNMONITORED / CLOSED transitions + neg-controls) pass;fleet.sh monitorsclassifies + ranks correctly by byproduct;shellcheck --severity=warningclean.🤖 Generated with Claude Code