菜单
Use this skill to coordinate concurrent agent work. Activates when tasks can be parallelized, during multi-agent workflows, or when orchestrating independent work streams.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | parallel-dispatch |
| description | Use this skill to coordinate concurrent agent work. Activates when tasks can be parallelized, during multi-agent workflows, or when orchestrating independent work streams. |
Coordinate multiple agents or work streams running concurrently.
Independent work should run in parallel.
If tasks don't depend on each other, running them sequentially wastes time.
┌─────────┐
│ START │
└────┬────┘
│
┌───────┴───────┐
▼ ▼
┌───────┐ ┌───────┐
│ Task A │ │ Task B │ ← Can run in parallel
└───┬───┘ └───┬───┘
│ │
└───────┬───────┘
▼
┌─────────┐
│ Task C │ ← Must wait for A and B
└────┬────┘
│
▼
┌─────────┐
│ END │
└─────────┘
tasks:
- id: A
depends_on: [] # Can start immediately
- id: B
depends_on: [] # Can start immediately
- id: C
depends_on: [A, B] # Must wait
- id: D
depends_on: [C] # Must wait for C
wave_1: [A, B] # Start together
wave_2: [C] # After wave_1 complete
wave_3: [D] # After wave_2 complete
## Dispatching Wave 1
Starting parallel tasks:
- [ ] Task A: [description] → Agent 1
- [ ] Task B: [description] → Agent 2
Waiting for completion...
## Wave 1 Results
Task A: ✅ Complete
- Output: [summary]
- Artifacts: [files changed]
Task B: ✅ Complete
- Output: [summary]
- Artifacts: [files changed]
Proceeding to Wave 2...
parallel_reviews:
- agent: security-engineer
focus: Security vulnerabilities
- agent: performance-engineer
focus: Performance issues
- agent: accessibility-auditor
focus: Accessibility compliance
aggregation:
combine: All findings into unified report
deduplicate: Similar issues
prioritize: By severity
parallel_changes:
- file: src/components/Button.tsx
change: Add loading state
- file: src/components/Input.tsx
change: Add error state
- file: src/components/Select.tsx
change: Add disabled state
merge_strategy: All changes are independent
parallel_tests:
- suite: unit
command: npm run test:unit
- suite: integration
command: npm run test:integration
- suite: e2e
command: npm run test:e2e
wait_for: All suites to complete
fail_fast: true # Stop all if one fails
agent_1_output:
status: complete
artifacts:
- src/components/Button.tsx
notes: Added loading prop, needs testing
agent_2_input:
receives: agent_1_output.artifacts
task: Write tests for loading state
conflict_detected:
file: src/lib/utils.ts
agent_1_change: Added formatDate function
agent_2_change: Added formatCurrency function
resolution:
strategy: merge_both # Both changes are compatible
# or: agent_1_wins / agent_2_wins / manual_review
## Parallel Execution Status
| Task | Agent | Status | Progress |
|------|-------|--------|----------|
| A | Agent-1 | 🔄 Running | 60% |
| B | Agent-2 | ✅ Complete | 100% |
| C | Agent-3 | ⏳ Waiting | 0% |
on_failure:
strategy: continue_others # Don't block parallel work
# or: fail_fast / retry
failed_task:
id: B
error: "TypeScript error in component"
action:
- Complete other parallel tasks
- Report failure in aggregation
- Decide: fix and retry, or proceed without
timeout:
per_task: 10 minutes
total_wave: 30 minutes
on_timeout:
- Cancel stuck task
- Log partial progress
- Continue with completed tasks
## Aggregated Results
### Completed Tasks
- Task A: Added Button component
- Task B: Added Input component
### Combined Changes
Files modified:
- src/components/Button.tsx (Task A)
- src/components/Input.tsx (Task B)
### Unified Test Results
- Unit tests: 45/45 passing
- Integration: 12/12 passing
### Issues Found
1. [Task A] Minor: Consider memo for Button
2. [Task B] Minor: Add aria-label to Input
Works with:
/riper - Execute phase can parallelize independent tasks/review - Parallel code review from multiple perspectivescode-review-workflow orchestrator - Multi-agent reviewWHEN to auto-invoke: Creating API routes, building endpoints, adding route.ts files, implementing REST/GraphQL APIs, adding authentication to APIs, rate limiting, API validation with Zod, handling HTTP methods (GET/POST/PUT/DELETE). WHEN NOT to invoke: Pure frontend work, styling, React components without API calls, documentation-only tasks, database migrations without API changes.
WHEN to auto-invoke: Code review requests, refactoring code, cleaning up code, performance optimization, security audits, linting issues, reducing technical debt. WHEN NOT to invoke: Building new features (use domain-specific skills), initial implementation, documentation-only work.
WHEN to auto-invoke: End of productive sessions, after solving difficult problems, after user corrections, after discovering effective workarounds, after establishing project conventions. WHEN NOT to invoke: Trivial sessions, sessions with only exploration, when explicitly told not to learn.
WHEN to auto-invoke: Database schema design, creating migrations, writing SQL queries, query optimization, Supabase operations, Prisma/Drizzle schema changes, PostgreSQL tasks, RLS policies, indexes. WHEN NOT to invoke: Pure frontend work, API logic without database, styling, React components, deployment tasks.
WHEN to auto-invoke: Creating CI/CD pipelines, GitHub Actions workflows, Docker/containerization, Vercel/Netlify deployment, infrastructure as code, environment variables, deployment configurations. WHEN NOT to invoke: Application code changes, UI development, database schema design, testing application logic.
WHEN to auto-invoke: Measuring code quality, testing AI outputs, establishing success criteria, defining acceptance tests, checking implementation correctness. WHEN NOT to invoke: Simple implementations, prototyping, exploration phases, when speed matters more than measurement.