brain-approve
Approve, reject, or manage pending brain memories. Use: /brain-approve [all|<id>]
Auto-learn project conventions from git history — zero manual setup. Trigger: /brain-learn
$ npx -y skills add Gr122lyBr/claude-teams-brain --skill brain-learn --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/brain-learnContext preview
The summary Claude sees to decide when to auto-load this skill.
Auto-learn project conventions from git history — zero manual setup. Trigger: /brain-learn
name: brain-learn description: | Auto-learn project conventions from git history — zero manual setup. Trigger: /brain-learn user_invocable: true
Scan the repo's git history and automatically extract conventions, architecture signals, file coupling patterns, and code hotspots. The brain teaches itself from your repo — no manual `/brain-remember` needed.
Run:
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/brain_engine.py learn "$CLAUDE_PROJECT_DIR"Parse the JSON response which contains:
Display results grouped by category:
## Learned from Git History **Commits analyzed**: 187 ### Conventions Added (6 new, 2 already known) - Convention: commit messages use Conventional Commits — common scopes: api, auth, db - Convention: branches use prefix naming (feature/, fix/, chore/) - Architecture: primary stack is TypeScript (Node.js) - Architecture: CI/CD uses GitHub Actions - Architecture: uses Docker for containerization - Convention: tests use *.test.ts naming ### Also Indexed - 12 file coupling patterns (searchable via `/brain-search coupling`) - 23 code hotspots (searchable via `/brain-search hotspots`)
After displaying results, suggest: > **Next steps:** > - `/brain-search coupling` — see which files change together > - `/brain-search hotspots` — see the most active files > - `/brain-status` — verify the brain state > - `/brain-forget <text>` — remove any convention that doesn't apply
The command is idempotent — running it again only adds net-new findings.
**Important**: Only use the exact command shown above. Do not invent or guess other command names.
Give your Claude Code Agent Teams a memory. Auto-injects role-specific context into every new teammate — your team never starts blind again.
Repo: Gr122lyBr/claude-teams-brain
Approve, reject, or manage pending brain memories. Use: /brain-approve [all|<id>]
Launch the brain dashboard web UI to visualize and curate memories, decisions, and files. Trigger: /brain-dashboard
Export accumulated brain knowledge as a CONVENTIONS.md file in your project. Distills all rules, decisions, files, and agent roles into a committable file.
Remove a manually stored memory from the brain by partial text match. Use: /brain-forget <text>
Export accumulated brain knowledge as `CONVENTIONS.md` and open a GitHub PR so your whole team can review what the AI team learned.