Repository navigation
feat(fleet): add /brain skill + brain/retire subcommands (v0.9.0) - #7
Merged
Merged
Conversation
Back-ports the brain (knowledge-oracle + Brain-agent seat) and retire (safe on-demand agent unload) capabilities into the SOURCE repo. They had shipped only in a downstream install and never landed here, so no fresh `install.sh` could pick them up. - /brain: a third agent skill under .fleet/skills/ (install.sh drops it into the target's .claude/skills/ like /coordinator + /arm). Rewritten GENERIC + leak-free for this PUBLIC repo — no private hub paths, no corpuscrawl, no agent numbers, no project specifics. Any memory-corpus tool is an OPTIONAL PATH integration; brain degrades gracefully when none is present and uses only Fleet's own DM / board / state. - fleet.sh brain <q> | ask "<q>" | serve | who | stand-down, plus fleet.sh retire [<agent>] [--force] — lazy-loaded from .fleet/bin/brain-retire.sh (hot path stays lean). - retire works for ANY agent (agent-N / short / session-id; no arg = self): releases claims, removes the record + wake breadcrumb, vacates the Brain seat if held, emits a retire board event. Safe-by-default — refuses a dirty-claim or stranded-DM retire without --force. - Forcing function: selftest-brain-retire.sh (hermetic, isolated state) proves the happy paths AND that the safety guards BITE; wired into CI as a dedicated selftest job run in a real git tree so the dirty-claim negative control actually fires (probe path moved out of .fleet/tmp/ so the repo's broad tmp/ gitignore can't silently mute it). Co-Authored-By: Claude Opus 4.8 (1M context) <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.
What & why
The brain (knowledge-oracle + Brain-agent seat) and retire (safe on-demand agent unload) capabilities had shipped only in a downstream install and were never back-ported to this source repo — so no fresh
install.shcould pick them up. This lands them here, sanitized for the public repo.Changes
/brain— a third agent skill under.fleet/skills/(installed into the target's.claude/skills/like/coordinator+/arm). Rewritten generic + leak-free: no private hub paths, nocorpuscrawl, no agent numbers, no project specifics. Any memory-corpus tool is an optional PATH integration;braindegrades gracefully when none is present and uses only Fleet's own DM / board / state.fleet.sh brain <q> | ask "<q>" | serve | who | stand-down+fleet.sh retire [<agent>] [--force]— lazy-loaded from.fleet/bin/brain-retire.shso the hot path stays lean.retireworks for ANY agent (agent-N/ short / session-id; no arg = self): releases claims, removes the record + wake breadcrumb, vacates the Brain seat if held, emits aretireboard event. Safe-by-default — refuses a dirty-claim or stranded-DM retire without--force.Forcing function
.fleet/bin/selftest-brain-retire.sh(hermetic, isolated state) proves the happy paths and that the safety guards BITE. Wired into CI as a dedicatedselftestjob run in a real git tree so the dirty-claim negative control actually fires — the probe path was moved out of.fleet/tmp/so the repo's broadtmp/gitignore can't silently mute it.Verification (byproduct)
shellcheck --severity=warningclean on all scriptsbrain/SKILL.mdinto.claude/skills/and ranbrain/retire(self + foreign agent by label)VERSION 0.8.1 → 0.9.0; CHANGELOG + README updated.
🤖 Generated with Claude Code