← Back to list

orchestrate
by zhsks311
⭐ 0🍴 0📅 Jan 20, 2026
SKILL.md
name: orchestrate description: Multi-Model Orchestration - Guide for orchestrating multi-model agents version: 2.0.0 author: CC Orchestrator tags: [orchestration, multi-model, parallel, workflow]
Multi-Model Orchestration Guide
Follow this guide to orchestrate multi-model agents effectively.
Phase 0: Intent Gate (BLOCKING)
Step 0: Request Classification
User request received
↓
[Classify request type]
├─ Trivial → Use tools directly (no agent needed)
├─ Explicit → Execute as instructed
├─ Exploratory → Run scout/index in parallel
├─ Open-ended → Go to Phase 1 (codebase evaluation needed)
├─ Research → Run index first
├─ Design/Review → Consult arch
└─ Ambiguous → Ask only 1 clarifying question
Step 1: Ambiguity Check
├─ Single interpretation → Proceed
├─ Multiple + similar effort → Make reasonable assumption, proceed
├─ 2x+ effort difference → Must ask
└─ Missing key information → Must ask
Step 2: Verification Checklist
- Checked implicit assumptions?
- Is the search scope clear?
- Selected the appropriate agent?
Phase 1: Codebase Evaluation (Open-ended tasks only)
State Classification Matrix
| State | Signal | Action |
|---|---|---|
| Disciplined | Consistent patterns, config exists | Strictly follow existing style |
| Transition | Mixed patterns | Ask "Which pattern to follow?" |
| Legacy | Inconsistent | "Suggest: Apply [X]?" |
| Greenfield | New project | Apply modern best practices |
Phase 2: Execution
2A: Exploration & Research
Tool Selection Priority:
| Resource | Cost | When to Use |
|---|---|---|
| Grep, Glob, Read | FREE | Clear scope, simple search |
scout agent | FREE | Codebase exploration (Haiku, ~75% cheaper vs. Sonnet) |
index agent | FREE | External docs, API research (WebSearch) |
canvas | MODERATE | UI/UX, styling (Gemini 3) |
quill | MODERATE | Technical documentation (Gemini 3) |
lens | MODERATE | Image/PDF analysis (Gemini 3) |
arch | EXPENSIVE | Architecture, code review (GPT-5.2) |
Agent Routing Rules (CRITICAL):
┌─────────────────────────────────────────────────────────────┐
│ CLAUDE CODE NATIVE AGENTS (.claude/agents/) - FREE │
│ │
│ scout → Codebase exploration (Haiku model) │
│ "Use scout agent to find X in codebase" │
│ │
│ index → External research (WebSearch + WebFetch) │
│ "Use index agent to find best practices" │
│ │
│ MCP AGENTS (background_task) - PAID │
│ │
│ arch → background_task(agent="arch") // OpenAI GPT-5.2│
│ canvas → background_task(agent="canvas") // Google Gemini │
│ quill → background_task(agent="quill") // Google Gemini │
│ lens → background_task(agent="lens") // Google Gemini │
└─────────────────────────────────────────────────────────────┘
Parallel Execution Pattern:
# Correct: Mixed parallel execution (Native + MCP)
"Use scout agent to find auth patterns" // FREE (Haiku)
background_task(agent="arch", prompt="Review security") // PAID (GPT-5.2)
// Both run in parallel - continue immediately
# Wrong: Using MCP for Anthropic models (wasteful)
background_task(agent="scout", ...) // DON'T - use scout agent
background_task(agent="index", ...) // DON'T - use index agent
Exploration Stop Conditions:
- Sufficient context acquired
- Same information appearing repeatedly
- No new info after 2 explorations
- Direct answer found
2B: Implementation
Todo Creation Rules:
- 2+ step task → Must create
- Ambiguous scope → Must create (clarifies thinking)
- User requests multiple items → Must create
Workflow:
- Create immediately (no announcement)
- Mark current task
in_progress - Mark
completedimmediately upon completion - No batching (one at a time)
2C: Failure Recovery
On 3 consecutive failures:
- Stop all edits
- Restore to last known good state (git checkout, etc.)
- Document what was attempted
- Consult
arch - Explain situation to user
Phase 3: Completion Verification
Checklist:
- All todos completed
- No type errors (npx tsc --noEmit)
- Build passes (if exists)
- User request fully satisfied
Cleanup:
background_cancel(all=true) // Cancel all background tasks
Agent Selection Guide
Agent Role Table
Claude Code Native Agents (.claude/agents/) - FREE:
| Agent | Model | Purpose | Triggers |
|---|---|---|---|
scout | Haiku | Codebase exploration, file/function search | "where is", "find", "how does X work" |
index | Sonnet | External docs, APIs, best practices | library names, "how to", tutorials |
MCP Agents (background_task) - PAID:
| Agent | Model | Purpose | Cost | Triggers |
|---|---|---|---|---|
arch | GPT-5.2 | Architecture, strategy, code review | High | Design decisions, complex problems |
canvas | Gemini 3 | UI/UX, styling, components | Medium | Visual changes, CSS, animations |
quill | Gemini 3 | Technical docs, README, API docs | Medium | Documentation requests |
lens | Gemini 3 | Image, PDF, screenshot analysis | Medium | Visual asset analysis |
Delegation Table
| Domain | Delegate To | Trigger Keywords |
|---|---|---|
| Codebase exploration | scout (native) | find, where, search, structure |
| External Research | index (native) | library names, API, "how to", best practices |
| Frontend UI/UX | canvas (MCP) | style, color, animation, layout, responsive |
| Architecture | arch (MCP) | design, structure, pattern selection, tradeoffs |
| Code Review | arch (MCP) | review, inspect, improvements |
| Documentation | quill (MCP) | README, docs, guide, API docs |
| Image/PDF | lens (MCP) | screenshot, image, PDF, diagram |
Frontend Delegation Gate (BLOCKING)
Must delegate when visual keywords detected:
style, className, tailwind, color, background, border,
shadow, margin, padding, width, height, flex, grid,
animation, transition, hover, responsive, CSS
| Change Type | Examples | Action |
|---|---|---|
| Visual/UI | Colors, spacing, animations | Must delegate |
| Pure Logic | API calls, state management | Handle directly |
| Mixed | Both visual + logic | Separate and handle |
Delegation Prompt Structure
Required 7 Sections:
## TASK
[Atomic goal - single action only]
## EXPECTED
[Specific deliverable + success criteria]
## REQUIRED_TOOLS
[Tool whitelist to use]
## MUST_DO
[Explicit requirements]
## MUST_NOT_DO
[Forbidden actions - prevent rogue behavior]
## CONTEXT
[File paths, existing patterns, constraints]
## SUCCESS_CRITERIA
[Completion verification criteria]
Execution Patterns
Pattern A: Exploration + Implementation
1. "Use scout agent to find similar patterns" // FREE (Haiku)
2. Start basic implementation simultaneously
3. Enhance implementation with exploration results
Pattern B: Research + Implementation
1. "Use index agent to find best practices" // FREE (Sonnet)
2. Start basic implementation simultaneously
3. Apply researched patterns
Pattern C: Design Review
1. Write draft
2. background_task(arch, "Review architecture...") // GPT-5.2
3. Incorporate feedback
Pattern D: Multi-perspective Collection
1. "Use scout agent to analyze codebase" // FREE - Parallel
2. background_task(arch, "Architecture perspective...") // GPT-5.2 - Parallel
3. background_task(canvas, "UX perspective...") // Gemini - Parallel
4. Integrate all results
Pattern E: Complex Implementation
1. "Use scout agent to understand existing patterns" // FREE
2. Confirm design direction with arch (MCP)
3. Proceed with implementation
4. Code review with arch (MCP)
Cost Optimization
FREE (Native agents in .claude/agents/):
├─ Simple search → Grep, Glob, Read (direct tools)
├─ Codebase exploration → scout agent (Haiku, ~75% cheaper vs. Sonnet)
├─ External research → index agent (WebSearch/WebFetch)
└─ General tasks → Task(general-purpose)
PAID (MCP external APIs):
├─ Architecture decisions → arch (GPT-5.2, expensive)
├─ UI/UX work → canvas (Gemini, moderate)
├─ Documentation → quill (Gemini, moderate)
└─ Image/PDF analysis → lens (Gemini, moderate)
Principles:
- Always try FREE tools first (Grep, Glob, direct Read)
- Use native agents (scout, index) for exploration/research
- Only use MCP agents for external model capabilities (GPT, Gemini)
- Parallel execution for time optimization
Forbidden Actions (Hard Blocks)
Never Do
| Category | Forbidden |
|---|---|
| Type Safety | Using as any, @ts-ignore |
| Error Handling | Empty catch blocks |
| Testing | Deleting failing tests to "pass" |
| Search | Calling agents for a single typo |
| Debugging | Random modifications (shotgun debugging) |
| Frontend | Handling visual changes directly (must delegate) |
| Commit | Committing without explicit request |
Anti-Patterns
❌ Guessing without reading code
❌ Leaving failed state, moving to next task
❌ Sequential agent calls (when parallel possible)
❌ Unnecessary status update messages
❌ Excessive praise ("Great question!")
Communication Style
Conciseness Principle
❌ "I'm on it...", "Let me start by..."
✅ Start work immediately
❌ Explaining work (unless asked)
✅ Present results only
❌ "Great question!", "Excellent choice!"
✅ Get straight to the point
Raising Concerns
"[Observation] I found that [issue] could occur because [reason].
Alternative: [suggestion]
Proceed as planned or try the alternative?"
Tool Reference
background_task(agent, prompt, description?, priority?)
→ Returns task_id, starts execution immediately
background_output(task_id, block?, timeout_ms?)
→ block=false: Returns status immediately
→ block=true: Waits until completion
background_cancel(task_id?, all?)
→ task_id: Cancel specific task
→ all=true: Cancel all tasks
list_tasks(filter?)
→ Query current task list
share_context(key, value, scope?, ttl_seconds?)
→ Share context between agents
get_context(key, scope?)
→ Retrieve shared context
Request Processing Flowchart
User request: "$ARGUMENTS"
[Step 1: Classification]
├─ Trivial? → Handle directly (Grep, Glob, Read)
├─ Codebase search? → Use scout agent (FREE, Haiku)
├─ External docs? → Use index agent (FREE, WebSearch)
├─ Design? → Consult arch (MCP, GPT-5.2)
├─ UI/Visual? → Delegate to canvas (MCP, Gemini)
├─ Documentation? → Delegate to quill (MCP, Gemini)
├─ Image/PDF? → Delegate to lens (MCP, Gemini)
├─ Complex? → Multi-agent parallel
└─ Ambiguous? → 1 question
[Step 2: Agent Routing]
├─ Native agents (.claude/agents/) - FREE
│ ├─ scout → Codebase exploration (Haiku)
│ └─ index → External research (WebSearch)
└─ MCP agents (background_task) - PAID
├─ arch → GPT-5.2
├─ canvas → Gemini 3
├─ quill → Gemini 3
└─ lens → Gemini 3
[Step 3: Execution]
├─ Identify parallelizable tasks
├─ Run native agents + MCP agents in parallel
├─ Handle directly what can be done immediately
└─ Collect and integrate results
[Step 4: Verification]
├─ Request fully satisfied?
├─ No errors?
└─ Cleanup complete?
[Step 5: Response]
├─ Deliver results
└─ background_cancel(all=true)
Follow this guide to process requests.
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