Meta-agent for orchestrating and prioritizing work across multiple agents.
Orchestrates multiple AI agents by aggregating their findings, prioritizing issues by severity and dependencies, and generating coordinated execution plans. Use when running multi-agent swarms to track progress, identify blockers, and recommend fix sequences.
/plugin marketplace add arevlo/claude-code-workflows/plugin install arevlo-swarm@claude-code-workflowsMeta-agent for orchestrating and prioritizing work across multiple agents.
Read findings from all running agents, prioritize issues, group related work, and provide actionable execution guidance. Acts as the "queen" in hive-mind orchestration.
Aggregate Findings
.claude/swarm/reports/Prioritize Work
Group Related Issues
Track Progress
Suggest Actions
# Read all recent reports
ls -t .claude/swarm/reports/*.md | head -20
Parse each report for:
For each issue:
- ID: unique identifier
- Source: which agent found it
- Severity: P0/HIGH/MEDIUM/LOW
- File: affected file path
- Line: line number(s)
- Description: brief issue description
- Fix: suggested resolution
- Status: pending/in_progress/fixed/skipped
For each file with issues:
- How many issues?
- Which agents flagged it?
- Does fixing it affect other files?
- Is it blocking other work?
Priority factors:
1. Severity (P0 > HIGH > MEDIUM > LOW)
2. Blocking status (blocking issues first)
3. Issue density (files with multiple issues)
4. Fix complexity (quick wins before complex)
## Coordinator Summary
**Generated:** <timestamp>
**Reports analyzed:** <count>
**Total issues:** <count>
### Priority Queue
| # | Priority | File | Issues | Agents | Action |
|---|----------|------|--------|--------|--------|
| 1 | P0 | src/api/auth.ts | 3 | silent-hunter, type-analyzer | Fix first (blocking) |
| 2 | P0 | src/handlers/user.ts | 2 | reviewer, silent-hunter | Fix second |
| 3 | HIGH | src/utils/validate.ts | 1 | type-analyzer | Quick fix |
| 4 | MEDIUM | src/components/Form.tsx | 2 | reviewer | Can parallelize |
### Grouped by File
**src/api/auth.ts** (3 issues, blocking)
| Agent | Issue | Suggested Fix |
|-------|-------|---------------|
| silent-hunter | Unhandled promise rejection L:45 | Add try-catch |
| silent-hunter | Missing error propagation L:67 | Rethrow or handle |
| type-analyzer | Unsafe type assertion L:52 | Add type guard |
**src/handlers/user.ts** (2 issues)
| Agent | Issue | Suggested Fix |
|-------|-------|---------------|
| reviewer | Function too complex L:30-80 | Extract helpers |
| silent-hunter | Uncaught async error L:55 | Add error boundary |
### Recommended Sequence
1. **Fix auth.ts first** (3 issues, blocks other handlers)
- Start with try-catch additions
- Then type guard
- Verify with existing tests
2. **Fix user.ts second** (2 issues, related to auth)
- Refactor complex function
- Add error handling
3. **Fix validate.ts** (1 issue, independent)
- Quick type fix
4. **Fix Form.tsx** (2 issues, UI layer)
- Can be done in parallel with backend fixes
### Progress Tracking
| Status | Count | Percentage |
|--------|-------|------------|
| Fixed | 0 | 0% |
| In Progress | 0 | 0% |
| Pending | 8 | 100% |
| Blocked | 0 | 0% |
### Blockers & Decisions Needed
- [ ] **Decision:** auth.ts L:45 - Should errors be logged or thrown? (Need human input)
- [ ] **Blocker:** No tests for auth module - fix blind or add tests first?
### Recommendations
1. **Immediate:** Fix P0 issues in auth.ts (security/stability risk)
2. **Today:** Complete HIGH priority items
3. **Consider:** Run test-analyzer if no tests exist for auth
4. **Watch:** Re-run silent-hunter after fixes to verify
### Agent Health
| Agent | Last Report | Issues Found | Status |
|-------|-------------|--------------|--------|
| reviewer | 5 min ago | 3 | Complete |
| type-analyzer | 5 min ago | 2 | Complete |
| silent-hunter | 5 min ago | 3 | Complete |
| Severity | Criteria | Score |
|---|---|---|
| P0 | Security, data loss, crashes | 100 |
| HIGH | Bugs, major smells, async errors | 75 |
| MEDIUM | Minor smells, type issues | 50 |
| LOW | Suggestions, style | 25 |
Bonus modifiers:
The coordinator agent is invoked by:
/hive --coordinate (future flag)/sync command (uses coordinator logic)/auto command (for multi-phase coordination)Trigger re-coordination when:
Use this agent when analyzing conversation transcripts to find behaviors worth preventing with hooks. Examples: <example>Context: User is running /hookify command without arguments user: "/hookify" assistant: "I'll analyze the conversation to find behaviors you want to prevent" <commentary>The /hookify command without arguments triggers conversation analysis to find unwanted behaviors.</commentary></example><example>Context: User wants to create hooks from recent frustrations user: "Can you look back at this conversation and help me create hooks for the mistakes you made?" assistant: "I'll use the conversation-analyzer agent to identify the issues and suggest hooks." <commentary>User explicitly asks to analyze conversation for mistakes that should be prevented.</commentary></example>