Agent skill
intent-router
Route user intent to optimal skill sequences. Use when the user request is ambiguous, requires multiple skills, or needs skill sequencing. Disambiguates intent, suggests skill workflows, and handles complex multi-step tasks. Integrates with testing, linting, and quality gates. Triggers on "which skill", "what to do", "sequence", "workflow", "ambiguous request", "combine skills".
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/skill-intent-router
SKILL.md
Intent Router
Purpose
Route ambiguous or complex user intents to optimal skill sequences. When a request could match multiple skills or requires multiple steps, this skill disambiguates and suggests the right workflow.
Problem Solved:
- "Fai tutti i check" → Which checks? In what order?
- "Testa tutto" → Unit? Integration? E2E?
- "Prepara il progetto" → Setup? Linting? Testing?
When to Use
- User request is ambiguous
- Multiple skills could apply
- Task requires skill sequencing
- Need to disambiguate intent
- Complex multi-step workflows
When NOT to use:
- Clear single-skill task
- Simple, direct request
- Already know which skill to use
Intent Classification
Intent Types
| Intent Type | Description | Example |
|---|---|---|
| Setup | Initialize project | "Setup this project" |
| Development | Write code with quality | "Implement feature X" |
| Validation | Check quality | "Run all checks" |
| Pre-PR | Pre-pull request | "Ready for PR" |
| Debug | Fix issues | "Something is broken" |
| Refactor | Improve existing code | "Clean up this code" |
Intent Patterns
INTENT_PATTERNS = {
'setup': [
r'setup|initialize|configure|bootstrap',
r'prepare.*project|get started|new project'
],
'development': [
r'implement|create|add|build|write',
r'feature|function|endpoint|component'
],
'validation': [
r'check|validate|verify|test|lint',
r'quality|coverage|all checks'
],
'pre_pr': [
r'ready.*pr|before.*merge|pre.*pr',
r'final check|quality gates|all green'
],
'debug': [
r'fix|debug|broken|error|issue|bug',
r'not working|fails|crash'
],
'refactor': [
r'refactor|clean.*up|improve|optimize',
r'restructure|simplify'
]
}
Skill Sequences
Sequence: Project Setup
User: "Setup this project"
Detected Intent: setup
Recommended Sequence:
1. [skill-test-setup] - Configure testing environment
2. [skill-linting-complete] - Setup linting
3. [skill-testing-philosophy] - Understand TDD approach
Execute? [Yes/No/Modify]
Sequence: Feature Development
User: "Implement user authentication"
Detected Intent: development
Recommended Sequence:
1. [skill-testing-philosophy] - TDD mindset
2. [skill-testing-workflow] - Choose testing tools
3. [skill-linting-complete] - Fix linting as you go
4. [skill-quality-gates] - Pre-PR validation
Execute? [Yes/No/Modify]
Sequence: Pre-PR Validation
User: "Run all checks before PR"
Detected Intent: pre_pr
Recommended Sequence:
1. [skill-quality-gates]
├─ Gate 1: [skill-linting-complete] - Linting
├─ Gate 2: Unit Tests
├─ Gate 3: Integration Tests
└─ Gate 4: [skill-testsprite-pre-pr] - E2E Tests
Execute? [Yes/No/Modify]
Sequence: Debug & Fix
User: "Tests are failing"
Detected Intent: debug
Recommended Sequence:
1. [skill-testing-workflow] - Analyze failures
2. [skill-testing-philosophy] - Fix with TDD
3. [skill-linting-complete] - Check linting
4. [skill-quality-gates] - Verify fix
Execute? [Yes/No/Modify]
Disambiguation Dialog
When Intent is Unclear
User: "Check everything"
Intent Router: I see you want to run checks, but I need to clarify:
What type of checks?
[1] Linting only (code style)
[2] Testing only (unit/integration)
[3] Full validation (lint + test + E2E)
[4] Pre-PR (all quality gates)
Please select [1-4] or describe what you need.
Multiple Possible Skills
User: "Test this API"
Intent Router: "Test API" could mean:
[1] Unit test the API client code
→ Use: [skill-testing-workflow]
[2] Integration test API endpoints
→ Use: [skill-testing-workflow] + integration config
[3] E2E test the API flow
→ Use: [skill-testsprite-pre-pr]
[4] All of the above
→ Sequence: [skill-testing-workflow] → [skill-testsprite-pre-pr]
Please select [1-4]
Skill Mapping Matrix
| User Says | Detected Intent | Skill Sequence |
|---|---|---|
| "Setup project" | setup | skill-test-setup → skill-linting-complete |
| "Create feature" | development | skill-testing-philosophy → skill-testing-workflow |
| "Run tests" | validation | skill-testing-workflow |
| "Check code quality" | validation | skill-linting-complete |
| "Ready for PR" | pre_pr | skill-quality-gates |
| "Fix bugs" | debug | skill-testing-workflow → skill-testing-philosophy |
| "Refactor code" | refactor | skill-linting-complete → skill-testing-workflow |
| "Optimize performance" | refactor | skill-testing-workflow → skill-testsprite-pre-pr |
| "All checks" | validation | skill-quality-gates |
| "Clean up" | refactor | skill-linting-complete |
Complex Workflow: Full Development Cycle
User: "Build a complete user authentication system"
Intent Router: This is a complex task requiring multiple skills.
Proposed Workflow:
PHASE 1: Setup
├─ [skill-test-setup] Configure testing
├─ [skill-linting-complete] Setup linting
└─ [skill-testing-philosophy] TDD approach
PHASE 2: Development
├─ [skill-testing-workflow] Unit tests first
├─ Implement auth logic
├─ [skill-linting-complete] Fix linting
└─ [skill-testing-workflow] Integration tests
PHASE 3: E2E Testing
├─ [skill-testsprite-pre-pr] User flow tests
└─ Login/logout/reset password flows
PHASE 4: Quality Gates
└─ [skill-quality-gates] Final validation
Execute full workflow? [Yes/Phase by Phase/Modify]
HITL Checkpoints
Human-in-the-Loop Decisions
Intent Router: Recommended sequence for "test everything":
1. [skill-linting-complete] - Check code style
2. [skill-testing-workflow] - Run unit tests
3. [skill-testsprite-pre-pr] - Run E2E tests
⏸️ CHECKPOINT: Review before execution
Options:
[✓] Execute all
[⏭️] Skip to specific step
[📝] Modify sequence
[❌] Cancel and clarify
Your choice?
Progress Updates
Executing sequence: "Ready for PR"
[✓] Gate 1: Linting - PASSED
[✓] Gate 2: Unit Tests - PASSED
[⏳] Gate 3: Integration Tests - RUNNING...
[⏸️] Gate 4: E2E Tests - PENDING
Continue? [Yes/Pause/Stop]
Intent Resolution Examples
Example 1: Ambiguous Request
User: "Make sure everything works"
Analysis:
- "everything" is vague
- "works" could mean: runs, passes tests, no lint errors
Disambiguation:
"What does 'works' mean to you?"
[1] Code compiles without errors
[2] All tests pass
[3] No linting issues
[4] Full quality gates (all of above)
User selects: [4]
Sequence: [skill-quality-gates]
Example 2: Implicit Intent
User: "I'm done with the feature"
Analysis:
- Implicit intent: ready to validate
- Missing: explicit request for checks
Inferred Intent: pre_pr
Suggested: "Run pre-PR quality gates?"
[Yes/No/What does that mean?]
Example 3: Context-Aware Routing
User: "Fix the tests"
Context: Previous conversation was about linting
Analysis:
- User might mean: fix linting issues OR fix test failures
- Context suggests: test failures
Sequence:
1. [skill-testing-workflow] - Analyze test failures
2. [skill-testing-philosophy] - Fix with TDD
3. [skill-quality-gates] - Verify all green
Fallback Strategies
Unknown Intent
User: "Do the thing"
Intent Router: I'm not sure what "the thing" refers to.
Could you clarify:
- Are you setting up something new?
- Are you checking existing code?
- Are you fixing an issue?
- Are you preparing for a PR?
Or describe what you want to accomplish.
No Matching Skills
User: "Deploy to production"
Intent Router: I don't have a skill for deployment.
Available skills cover:
✓ Testing and quality
✓ Linting and code style
✓ Development workflows
For deployment, you might need:
- CI/CD configuration
- Infrastructure setup
- Deployment scripts
Should I help you create a deployment skill?
Integration with Other Skills
| Skill | Integration |
|---|---|
| skill-smart-loading | Loads skills identified by intent-router |
| skill-model-routing | Selects model for execution |
| skill-quality-gates | Final validation in sequences |
| All testing skills | Part of validation sequences |
| All linting skills | Part of quality sequences |
Best Practices
Do
- ✅ Ask for clarification when ambiguous
- ✅ Suggest sequences, don't force them
- ✅ Explain WHY a sequence is recommended
- ✅ Allow modification of sequences
- ✅ Remember context from previous interactions
Don't
- ❌ Guess when highly uncertain
- ❌ Execute long sequences without checkpoints
- ❌ Ignore user corrections
- ❌ Assume intent without confirmation
- ❌ Overwhelm with too many options
Learning from Interactions
class IntentLearner:
"""
Learn from routing decisions
"""
def __init__(self):
self.routing_history = []
def log_routing(self, user_input, detected_intent, suggested_sequence, user_accepted):
"""
Log routing for learning
"""
self.routing_history.append({
'input': user_input,
'intent': detected_intent,
'sequence': suggested_sequence,
'accepted': user_accepted,
'timestamp': datetime.now()
})
def improve_patterns(self):
"""
Update intent patterns based on history
"""
# Analyze which patterns led to accepted routes
# Update INTENT_PATTERNS accordingly
pass
Version
v1.0.0 (2025-01-28) - Intent routing with skill sequencing
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?