home / skills / bbeierle12 / skill-mcp-claude / router-template
This skill routes requests to multiple specialized skills by classifying, matching, and loading the best ensembles for production tasks.
npx playbooks add skill bbeierle12/skill-mcp-claude --skill router-templateReview the files below or copy the command above to add this skill to your agents.
---
name: router-template
description: Template for creating router skills. Use as a starting point when building a new router that dispatches to multiple specialized skills. Copy and customize for your domain.
---
# [Domain] Router
Routes to N specialized skills based on task requirements.
## Routing Protocol
1. **Classify** — Identify primary task type from user request
2. **Match** — Find skill(s) with highest signal match
3. **Combine** — Most production tasks need 2-4 skills together
4. **Load** — Read matched SKILL.md files before implementation
## Quick Route
### Tier 1: Core (Always Consider)
| Task Type | Skill | Primary Signal Words |
|-----------|-------|---------------------|
| [task] | `skill-name` | word1, word2, word3 |
### Tier 2: Enhanced (Add When Needed)
| Task Type | Skill | Primary Signal Words |
|-----------|-------|---------------------|
| [task] | `skill-name` | word1, word2, word3 |
### Tier 3: Reference Only
| Task Type | Skill | Primary Signal Words |
|-----------|-------|---------------------|
| [task] | `skill-name` | word1, word2, word3 |
## Signal Matching Rules
### Priority Order
When multiple signals present, resolve by priority:
1. **Explicit framework** — React/Vue/vanilla overrides generic signals
2. **Specific feature** — "multi-step wizard" beats generic "form"
3. **Platform constraint** — mobile/VR/accessibility takes precedence
4. **Default tier** — Fall back to Tier 1 core skills
### Confidence Scoring
- **High (3+ signals)** — Route immediately
- **Medium (1-2 signals)** — Route with Tier 1 defaults
- **Low (0 signals)** — Ask user for clarification
## Common Combinations
### [Scenario Name] (N skills)
```
skill-1 → [what it provides]
skill-2 → [what it provides]
skill-3 → [what it provides]
```
Wiring: [one-line integration summary]
### [Scenario Name] (N skills)
```
skill-1 → [what it provides]
skill-2 → [what it provides]
```
Wiring: [one-line integration summary]
## Decision Table
| Framework | Scale | Features | Route To |
|-----------|-------|----------|----------|
| React | Standard | Auth form | skill-a + skill-b + skill-c |
| Vue | Standard | Data entry | skill-a + skill-d |
| None | Simple | Basic form | skill-e + skill-b |
| Any | Complex | Wizard | skill-f + skill-a + skill-b |
## Fallback Behavior
- **Unknown framework** → Use skill with broadest compatibility
- **No clear signals** → Ask: "What framework?" and "What's the core requirement?"
- **Conflicting signals** → Prefer Tier 1 skills, ask for clarification on Tier 2
## Reference
See `references/integration-guide.md` for:
- Complete wiring patterns
- Code examples for all combinations
- Architecture diagrams
- Edge case handling
This skill is a router template that dispatches requests to multiple specialized skills. It provides a lightweight protocol for classifying requests, matching relevant skills, and combining them into a multi-skill workflow. Use it as a starting point to build a domain-specific router that coordinates 2–4 skills per task.
The router classifies the user request to identify the primary task type, then scores available skills by signal match. It applies priority rules to resolve conflicting signals, selects the highest-confidence skills, and recommends combinations for production tasks. When signals are weak, the router prompts for clarification to improve routing accuracy.
How many skills should be combined for a typical production task?
Most production tasks combine 2–4 skills: one for core functionality, one or two for features, and one for integration or platform concerns.
What if multiple signals conflict?
Resolve by priority: explicit framework > specific feature > platform constraint > default tier. If still ambiguous, prefer core skills and ask the user a clarifying question.