feature-council
Multi-agent feature implementation. Spawns independent solver agents that each implement the feature from scratch, then synthesizes the best elements from each. Use when building complex features where you want diverse approaches and comprehensive edge case coverage.
michaelboeding · 已有静态检查记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add michaelboeding/skills --skill feature-council技能说明
name: feature-council description: Multi-agent feature implementation. Spawns independent solver agents that each implement the feature from scratch, then synthesizes the best elements from each. Use when building complex features where you want diverse approaches and comprehensive edge case coverage.
Feature Council: Multi-Agent Feature Implementation
Spawns multiple agents to implement the same feature independently, then synthesizes the best parts from each. Unlike code-council (majority voting for bugs), this skill merges strengths because features have multiple valid approaches.
Use this for complex features where you want diverse implementations and comprehensive coverage.
Step 0: Ask User How Many Agents
Before doing anything else, ask the user how many solver agents to use:
How many solver agents would you like me to use? (3-10)
Recommendations:
- 3 agents: Faster, good for straightforward features
- 5 agents: Good balance of diversity and speed
- 7 agents: Comprehensive coverage
- 10 agents: Maximum diversity (complex features)
Note: Each agent will independently implement the entire feature.
More agents = more diverse approaches and edge case coverage.
Wait for the user's response. If they specified a number (e.g., "feature council of 5"), use that.
Minimum: 3 agents | Maximum: 10 agents
CRITICAL: Pure Independence
What This Means
- NO orchestrator exploration - Do NOT read files or gather context before spawning agents
- Raw user prompt to all agents - Each agent gets the user's original request, unchanged
- Each agent explores independently - Agents discover the codebase themselves
- Each agent implements fully - Complete, working implementations
Why This Matters
Independent implementations mean:
- Different architectural choices to compare
- Different edge cases discovered
- Different error handling approaches
- More comprehensive final solution
Workflow
Step 1: Capture the Raw User Prompt
Take the user's request exactly as stated. Do NOT:
- ❌ Read files first
- ❌ Explore the codebase
- ❌ Add context
- ❌ Rephrase or enhance the prompt
Just capture what the user said.
Step 2: Spawn Agents IN PARALLEL with RAW PROMPT
Spawn ALL agents simultaneously. Each gets the exact same raw prompt:
Task(agent: "feature-solver-1", prompt: "[USER'S EXACT WORDS]")
Task(agent: "feature-solver-2", prompt: "[USER'S EXACT WORDS]")
Task(agent: "feature-solver-3", prompt: "[USER'S EXACT WORDS]")
... (all in the SAME batch - parallel execution)
DO NOT modify the prompt. DO NOT add context. Raw user words only.
Step 3: Agents Work Independently
Each agent will:
- Read and understand the user's feature request
- Explore the codebase to understand existing patterns
- Design their approach
- Implement the complete feature
- Handle edge cases they identify
- Verify their implementation
Each agent works in complete isolation.
Step 4: Track Progress & Collect Solutions
As agents complete, show progress to the user:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
AGENT PROGRESS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
☑ Agent 1 - Complete
☑ Agent 2 - Complete
☑ Agent 3 - Complete
☐ Agent 4 - Working...
☐ Agent 5 - Working...
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Collect all outputs when complete.
Step 5: Analyze & Compare Implementations
DO NOT use majority voting. Instead, analyze each implementation for:
| Category | What to Look For |
|---|---|
| Architecture | Design patterns, code organization, modularity |
| Edge Cases | What edge cases did each agent handle? |
| Error Handling | How robust is the error handling? |
| Type Safety | Type definitions, null checks, validation |
| Performance | Efficiency, caching, optimization |
| Maintainability | Readability, documentation, testability |
| Codebase Fit | How well does it match existing patterns? |
Create a comparison matrix:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
IMPLEMENTATION COMPARISON
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
| Aspect | Agent 1 | Agent 2 | Agent 3 | Agent 4 | Agent 5 |
|-----------------|---------|---------|---------|---------|---------|
| Architecture | MVC | Service | MVC | MVC | Modular |
| Edge Cases | 3 | 5 | 4 | 3 | 6 |
| Error Handling | Basic | Robust | Good | Basic | Robust |
| Type Safety | ✓ | ✓✓ | ✓ | ✓ | ✓✓ |
| Codebase Match | 90% | 75% | 95% | 85% | 80% |
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Step 6: Synthesize Best Solution
Combine the best elements from each agent:
- Select base implementation - Choose the one that best matches codebase patterns
- Incorporate edge cases - Add edge cases from other agents the base missed
- Enhance error handling - Use the most robust error handling approach
- Improve type safety - Merge type definitions and validations
- Document sources - Track which elements came from which agent
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SYNTHESIS DECISION
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Base: Agent 3 (best codebase pattern match)
Incorporating from other agents:
├─ Agent 2: Robust error handling pattern
├─ Agent 5: Edge cases for [empty input, concurrent access]
├─ Agent 2: Type definitions for [UserInput, ValidationResult]
└─ Agent 4: Caching optimization
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Step 7: Create Finalized Implementation Plan
Before implementing, create a detailed execution plan from the synthesized solution:
━━━━━
阅读完整来源 ↗