← Back to list

agent-coordination
by pagerguild
Development environment automation with multi-agent workflow orchestration for Claude Code
⭐ 0🍴 0📅 Jan 16, 2026
SKILL.md
name: agent-coordination description: | Use when coordinating multiple AI agents for parallel work. Triggers: "multi-agent", "parallel agents", "agent coordination", "QuantumDAG", "conflict-free". NOT for: Single-agent operations or basic jj commands.
Agent Coordination
Expert guidance for multi-agent coordination using QuantumDAG.
Core Concepts
QuantumDAG
A Directed Acyclic Graph structure for tracking agent operations without conflicts:
- Conflict-free - Multiple agents can work simultaneously
- Automatic detection - Conflicts detected before they occur
- 23x faster - Than Git for 8-10 agent scenarios
- Quantum fingerprints - SHA3-512 for integrity verification
Agent Lifecycle
┌─────────────────────────────────────────────┐
│ │
│ Register → Check Conflicts → Operate → │
│ ↑ │ │
│ └──────── Report ←─────────────┘ │
│ │
└─────────────────────────────────────────────┘
Coordination API
Setup
const { JjWrapper } = require('agentic-jujutsu');
const jj = new JjWrapper();
await jj.enableAgentCoordination();
Register Agent
// Each agent must register before coordinating
await jj.registerAgent('agent-001', 'code-reviewer');
await jj.registerAgent('agent-002', 'test-automator');
await jj.registerAgent('agent-003', 'security-auditor');
Check Conflicts Before Operating
// ALWAYS check before modifying files
const conflicts = await jj.checkAgentConflicts(
'operation-id',
'edit', // Operation type: 'edit', 'create', 'delete'
['src/auth.ts'] // Files you plan to touch
);
if (conflicts.length > 0) {
// Another agent is working on these files
console.log('Wait for:', conflicts.map(c => c.agentId));
return;
}
// Safe to proceed
await jj.registerAgentOperation('agent-001', 'operation-id', ['src/auth.ts']);
Get Coordination Stats
const stats = await jj.getCoordinationStats();
// {
// totalAgents: 3,
// activeOperations: 2,
// conflictsDetected: 0,
// averageResolutionTime: 45
// }
Workflow Patterns
Pattern 1: Parallel Code Review
// Launch 3 reviewers in parallel
const reviewers = ['code-reviewer', 'security-auditor', 'architect-reviewer'];
for (const type of reviewers) {
await jj.registerAgent(`${type}-${Date.now()}`, type);
}
// Each reviewer checks conflicts before reading
// No conflicts for read-only operations
Pattern 2: Parallel File Editing
// Agent A wants to edit auth.ts
const conflictsA = await jj.checkAgentConflicts('op-A', 'edit', ['auth.ts']);
// Agent B wants to edit user.ts (different file - no conflict)
const conflictsB = await jj.checkAgentConflicts('op-B', 'edit', ['user.ts']);
// Both can proceed in parallel
Pattern 3: Sequential with Handoff
// Agent A completes work
await jj.registerAgentOperation('agent-A', 'op-1', ['shared.ts']);
// ... do work ...
// Operation complete - automatically released
// Agent B can now work on same file
const conflicts = await jj.checkAgentConflicts('op-2', 'edit', ['shared.ts']);
// conflicts = [] (empty, A is done)
Best Practices
1. Always Register First
// DO
await jj.enableAgentCoordination();
await jj.registerAgent(agentId, agentType);
// DON'T
// Skip registration and hope for the best
2. Check Before Write
// DO
const conflicts = await jj.checkAgentConflicts(opId, 'edit', files);
if (conflicts.length === 0) {
// Safe to edit
}
// DON'T
// Edit files without checking
3. Use Specific File Lists
// DO
await jj.checkAgentConflicts(opId, 'edit', ['src/auth.ts', 'src/user.ts']);
// DON'T
await jj.checkAgentConflicts(opId, 'edit', ['src/']); // Too broad
4. Handle Conflicts Gracefully
if (conflicts.length > 0) {
// Option 1: Wait
await sleep(1000);
return retry();
// Option 2: Work on different files
const safeFiles = files.filter(f => !conflicts.some(c => c.files.includes(f)));
// Option 3: Report to orchestrator
throw new ConflictError(conflicts);
}
Performance Characteristics
| Agents | QuantumDAG | Git | Speedup |
|---|---|---|---|
| 2 | 5ms | 12ms | 2.4x |
| 4 | 8ms | 35ms | 4.4x |
| 8 | 15ms | 180ms | 12x |
| 10 | 20ms | 450ms | 23x |
Troubleshooting
"Coordination not enabled"
// Always enable before using coordination features
await jj.enableAgentCoordination();
"Agent not registered"
// Register before operations
await jj.registerAgent(agentId, agentType);
"Stale conflict data"
// Get fresh tips
const tips = await jj.getCoordinationTips();
Related
/agentic-flow- Command interfaceagentsdb-patterns- Learning from operationsquantum-signing- Operation integritydocs/JJ-INTEGRATION.md- Full API reference
Score
Total Score
50/100
Based on repository quality metrics
✓SKILL.md
SKILL.mdファイルが含まれている
+20
○LICENSE
ライセンスが設定されている
0/10
○説明文
100文字以上の説明がある
0/10
○人気
GitHub Stars 100以上
0/15
○最近の活動
3ヶ月以内に更新がある
0/10
○フォーク
10回以上フォークされている
0/5
✓Issue管理
オープンIssueが50未満
+5
✓言語
プログラミング言語が設定されている
+5
○タグ
1つ以上のタグが設定されている
0/5
Reviews
💬
Reviews coming soon