Skip to content

LeoYeAI/agent-master-skills / skills/ai-model-router-v2/skill

FileKindSize
SKILL.mdskill md3.2 KB
core/router.pyscript13.3 KB
modules/context.pyscript3.1 KB
modules/detector.pyscript3.6 KB
---
name: ai-model-router
description: Intelligent AI model router that automatically switches between two configured models (local for simple tasks, cloud for complex ones). Detects local models (Ollama, LM Studio) automatically, routes based on task complexity and privacy. Use when users ask to "switch model", "use local/cloud model", or mention API keys/passwords (triggers privacy mode). Trigger on: sensitive data detection (forces local), complex tasks like "design architecture" (uses cloud), or configuration requests.
version: 1.0.0
---

# AI Model Router

Compact, intelligent model routing that just works.

## Quick Start

```bash
# Install
npx clawhub@latest install ai-model-router

# First run - auto-detects your models
python3 skill/core/router.py "What is Python?"

# List available models
python3 skill/core/router.py --list
```

## How It Works

```
Your Request → Analyze → Select Model
                          ↓
                    Simple? → Primary (fast/cheap)
                    Complex? → Secondary (capable)
                    Private? → Primary (forced)
```

## Scoring (from model-router-premium)

| Pattern | Points |
|---------|--------|
| Microservices, architecture | +10 |
| Design, implement, optimize | +5 |
| Explain, analyze, compare | +3 |
| **Syntax, example, "what is"** | **-3** |

**Threshold: 5** (simple vs complex)

## Features

| Feature | Status |
|---------|--------|
| Auto-detect local models | ✓ (Ollama, LM Studio) |
| Cloud model registry | ✓ (7 built-in) |
| Privacy detection | ✓ (API keys, passwords) |
| Context tracking | ✓ (conversations) |
| JSON config | ✓ (optional) |
| CLI interface | ✓ |
| **Core code size** | **~200 lines** |

## CLI

```bash
# Route a task
python3 skill/core/router.py "Design a system"
python3 skill/core/router.py "What is a for loop?"

# Options
--json                    # JSON output
--force primary           # Force primary model
--list                    # List all models
--status                  # Show status
```

## Python API

```python
from skill.core.router import RouterCore

router = RouterCore()
result = router.route("Design microservices")

print(result.model_name)   # "Claude Opus 4"
print(result.reason)        # "complex_task(score=15)"
print(result.confidence)    # 0.75
```

## Configuration (Optional)

Create `~/.model-router/models.json`:

```json
{
  "primary_model": {"id": "ollama:llama3:8b"},
  "secondary_model": {"id": "anthropic:claude-opus-4"},
  "models": [...]
}
```

**Without config**: Auto-detects local + uses cloud registry.

## Privacy Protection

Automatically forces primary (local) when sensitive data detected:
- API keys (`sk-...`, `api_key`)
- Passwords (`password`, `passwd`)
- Tokens (`bearer`, `secret`)
- Emails, SSN, credit cards

## Files

- `core/router.py` - Core routing engine (~200 lines)
- `modules/detector.py` - Auto-detection (optional)
- `modules/context.py` - Context tracking (optional)

## Inspired By

- **model-router-premium**: Simple scoring logic, cost-aware routing
- **Model Router v1**: Full feature set, documentation

This version combines:
- The **simplicity** of model-router-premium (~200 lines)
- The **features** of ai-model-router (privacy, auto-detect, context)
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall