Intelligent Routing

Intelligent Agent Routing

> Purpose: Automatically analyze user requests and route them to the most appropriate specialist agent(s) without requiring explicit user mentions.

Core Principle

> The AI should act as an intelligent Project Manager, analyzing each request and automatically selecting the best specialist(s) for the job while maintaining Trust-Grade governance.

---

How It Works

1. Request Analysis

Before responding to ANY user request, perform automatic analysis:

graph TD
    A[User Request] --> B[ANALYZE]
    B --> C[Keywords]
    B --> D[Domains]
    B --> E[Complexity]
    C --> F[SELECT AGENT]
    D --> F
    E --> F
    F --> G[Apply Trust-Grade Context]
    G --> H[AUTO-INVOKE with governance]

2. Agent Selection Matrix

| User Intent | Keywords | Selected Agent(s) | Auto-invoke? | | ----------------- | ----------------------------------- | -------------------------- | ------------ | | Architecture | "design", "structure", "pattern" | architect | ✅ YES | | Planning | "plan", "roadmap", "sprint" | planner | ✅ YES | | Code Review | "review", "check", "audit" | code-reviewer | ✅ YES | | Security | "security", "vulnerability", "auth" | security-reviewer | ✅ YES | | Testing | "test", "coverage", "e2e" | tdd-guide + e2e-runner | ✅ YES | | Build Errors | "error", "build", "compile" | build-error-resolver | ✅ YES | | Refactoring | "refactor", "clean", "improve" | refactor-cleaner | ✅ YES | | Documentation | "docs", "readme", "document" | doc-updater | ✅ YES | | Knowledge | "learn", "remember", "pattern" | knowledge-agent | ✅ YES | | Complex Task | Multiple domains detected | planner → multi-agent | ⚠️ ASK FIRST |

---

Domain Detection Rules

Single-Domain Tasks (Auto-invoke Single Agent)

| Domain | Patterns | Agent | | ---------------- | ------------------------------------- | -------------------------- | | Architecture | design, pattern, structure, layer | architect | | Planning | plan, task, sprint, milestone | planner | | Security | auth, jwt, password, vulnerability | security-reviewer | | Testing | test, jest, coverage, e2e, playwright | tdd-guide / e2e-runner | | Build | error, compile, typescript, lint | build-error-resolver | | Refactor | clean, refactor, improve, optimize | refactor-cleaner | | Docs | readme, document, api-docs | doc-updater |

Multi-Domain Tasks (Orchestration Required)

If request matches 2+ domains from different categories, escalate to planner for orchestration:

Example: "Create a secure login with tests"
→ Detected: Security + Testing + Backend
→ Auto-invoke: planner (orchestration mode)
→ Planner will handle: security-reviewer, architect, tdd-guide

---

Complexity Assessment

SIMPLE (Direct agent invocation)

• Single file edit
• Clear, specific task
• One domain only
• Example: "Fix the TypeScript error in auth.ts"

Action: Auto-invoke respective agent

MODERATE (2-3 agents)

• 2-3 files affected
• Clear requirements
• 2 domains max
• Example: "Add API endpoint with tests"

Action: Auto-invoke relevant agents sequentially

COMPLEX (Orchestration required)

• Multiple files/domains
• Architectural decisions needed
• Unclear requirements
• Example: "Build a new feature vertical"

Action: Auto-invoke planner → will ask Socratic questions

---

Implementation Rules

Rule 1: Silent Analysis

DO NOT announce "I'm analyzing your request..."

• ✅ Analyze silently
• ✅ Inform which agent is being applied
• ❌ Avoid verbose meta-commentary

Rule 2: Inform Agent Selection

DO inform which expertise is being applied:

🤖 **Applying `@security-reviewer` expertise...**

I will review the authentication implementation with the following focus:
[Continue with specialized response]

Rule 3: Trust-Grade Context

Always apply project governance context:

• Load relevant governance rules
• Apply session state context
• Enforce pre-task checklists

Rule 4: Override Capability

User can still explicitly mention agents:

User: "Use @architect to review this"
→ Override auto-selection
→ Use explicitly mentioned agent

---

Integration with Project Governance

With Session State

• Check session-state.json for active context
• Inherit project-specific configurations
• Maintain Trust-Grade continuity

With Governance Rules

• Apply relevant governance rules
• Enforce professional standards
• Maintain project architecture alignment

With Checklists

• Trigger pre-task checks for complex operations
• Validate against session-end protocols
• Ensure documentation updates

---

Summary

1. Analyze every request before responding
2. Select best agent(s) using the matrix
3. Inform user which expertise is applied
4. Maintain governance through Trust-Grade protocols
5. Stay seamless - user shouldn't notice the routing

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: besync-labs
• Source: besync-labs/antigravity-ai-kit
• License: MIT
• Homepage: https://besync-labs.github.io/antigravity-ai-kit/

Install and usage instructions live in the source repository linked above.