Claude-skill-registry dev-router
Developer agent skill router - routes to 31 dev skills by category and signal keywords, manages 5 sub-agents.
git clone https://github.com/majiayu000/claude-skill-registry
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/dev-router" ~/.claude/skills/majiayu000-claude-skill-registry-dev-router && rm -rf "$T"
skills/data/dev-router/SKILL.mdDeveloper Skill Router
"Right skill, right sub-agent, right time."
Common Skill Combinations
Typical task types and the skill combinations that work well for them.
Task Type → Skill Combination Mapping
1. New Game Feature (e.g., Player Movement)
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-r3f-r3f-fundamentals
Why: Research ensures patterns are followed.
2. Multiplayer Feature (e.g., Room Creation)
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-multiplayer-server-authoritative
- dev-multiplayer-colyseus-server
- dev-multiplayer-colyseus-state
- dev-multiplayer-colyseus-client
Why: Full server-authoritative stack from architecture to client. Validation ensures server builds correctly.
3. Client-Side Prediction (e.g., Movement Prediction)
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY )
- dev-multiplayer-server-authoritative
- dev-multiplayer-prediction-basics
- dev-multiplayer-prediction-movement
- dev-typescript-advanced
Why: Prediction builds on server-authoritative foundation. Advanced TypeScript for generic prediction types.
4. Asset Loading (e.g., Character Models)
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-assets-vite-asset-loading
- dev-assets-model-loading
- dev-r3f-r3f-fundamentals
Why: Vite patterns + specific model loading. R3F fundamentals for scene integration.
5. Performance Optimization (e.g., FPS Drop)
Skills:
- dev-research-codebase-exploration (MANDATORY)
- dev-performance-performance-basics
- dev-performance-instancing (if many objects)
- dev-performance-lod-systems (if complex models)
- dev-patterns-object-pooling (if transient objects)
Why: Start with basics, add specific techniques based on problem. Object pooling for per-frame creation.
6. UI/HUD Implementation
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-typescript-basics
- dev-patterns-ui-animations
Why: TypeScript for component props. UI animations for polish. E2E testing for UI validation.
7. Shooting Mechanics
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-multiplayer-server-authoritative
- dev-multiplayer-prediction-shooting
- dev-r3f-r3f-fundamentals
- dev-patterns-object-pooling
Why: Server-authoritative hit detection. Shooting prediction. Object pooling for projectiles.
8. Territory/Coverage System
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-patterns-coverage-tracking
- dev-typescript-advanced
- dev-r3f-r3f-fundamentals
Why: Coverage tracking pattern. Advanced TypeScript for grid state types. R3F for visualization.
9. Mobile Touch Controls
Skills:
- dev-research-gdd-reading (MANDATORY)
- dev-research-codebase-exploration (MANDATORY)
- dev-performance-mobile-optimization
- dev-patterns-mobile-haptics
- dev-typescript-basics
Why: Mobile optimization patterns. Haptics for feedback. E2E testing with touch events.
10. Bug Fix (any category)
Skills:
- dev-research-pattern-finding
- [Category-specific skill based on bug]
Why: Find existing patterns first. Load only the relevant domain skill. Validation to ensure fix works.
Skill Load Order
For complex tasks, load skills in this order:
- Research first: dev-research-codebase-exploration (always)
- Foundations: r3f-fundamentals, typescript-basics, server-authoritative
- Domain specific: physics, materials, prediction, etc.
- Patterns: object-pooling, ui-animations
- Validation last: feedback-loops, browser-testing
Combining with Sub-Agents
Task({ subagent_type: "developer-code-research", ... }) ↓ [Research completes, provides findings] ↓ Load domain skills based on findings ↓ Task({ subagent_type: "code-implementation", ... }) ↓ [Implementation completes] ↓ Task({ subagent_type: "developer-validation", ... }) ↓ [Validation passes] ↓ Task({ subagent_type: "commit-agent", ... })
Anti-Patterns to Avoid
| Anti-Pattern | Why | Correct Approach |
|---|---|---|
| Loading all R3F skills at once | Context bloat | Load only relevant skills |
| Loading validation before research | Wasted cycles | Research → Implement → Validate |
| Skipping research for "simple" tasks | Miss patterns | Always research first |
| Loading both basics and advanced TypeScript | Redundant | Advanced includes basics concepts |
| Loading multiplayer for single-player features | Unused code | Only load when needed |