admin/QwenClaw-with-Auth
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
c700d84f4f315539532417d1bb1846631de5769a
QwenClaw-with-Auth/docs/RIG-INTEGRATION.md

8.7 KiB
Raw Blame History

QwenClaw Rig Integration

Overview

QwenClaw now integrates with Rig (https://github.com/0xPlaygrounds/rig), a high-performance Rust AI agent framework, providing:

  • 🤖 Multi-Agent Orchestration - Agent councils for complex tasks
  • 🛠️ Dynamic Tool Calling - Context-aware tool resolution
  • 📚 RAG Workflows - Vector store integration for semantic search
  • ⚡ High Performance - Rust-native speed and efficiency

Architecture

┌─────────────────┐     ┌─────────────────┐
│  QwenClaw       │────▶│  Rig Service    │
│  (TypeScript)   │◀────│  (Rust + Rig)   │
│  - Daemon       │     │  - Agents       │
│  - Web UI       │     │  - Tools        │
│  - Scheduling   │     │  - Vector Store │
└─────────────────┘     └─────────────────┘
         │                       │
         ▼                       ▼
    Qwen Code              OpenAI/Anthropic
    Telegram               SQLite Vectors

Quick Start

1. Start Rig Service

cd rig-service

# Set environment variables
export OPENAI_API_KEY="your-key-here"
export RIG_HOST="127.0.0.1"
export RIG_PORT="8080"

# Build and run
cargo build --release
cargo run

2. Use Rig in QwenClaw

import { initRigClient, executeWithRig } from "./src/rig";

// Initialize Rig client
const rig = initRigClient("127.0.0.1", 8080);

// Check if Rig is available
if (await rig.health()) {
  console.log("✅ Rig service is running!");
}

// Create an agent
const sessionId = await rig.createAgent({
  name: "researcher",
  preamble: "You are a research specialist.",
  model: "gpt-4",
});

// Execute prompt
const result = await executeWithRig(sessionId, "Research AI trends in 2026");
console.log(result);

API Reference

Agents

// Create agent
const sessionId = await rig.createAgent({
  name: "assistant",
  preamble: "You are a helpful assistant.",
  model: "gpt-4",
  provider: "openai",
  temperature: 0.7,
});

// List agents
const agents = await rig.listAgents();

// Execute prompt
const response = await rig.executePrompt(sessionId, "Hello!");

// Get agent details
const agent = await rig.getAgent(sessionId);

// Delete agent
await rig.deleteAgent(sessionId);

Multi-Agent Councils

// Create council with multiple agents
const councilId = await rig.createCouncil("Research Team", [
  {
    name: "researcher",
    preamble: "You are a research specialist.",
    model: "gpt-4",
  },
  {
    name: "analyst",
    preamble: "You are a data analyst.",
    model: "gpt-4",
  },
  {
    name: "writer",
    preamble: "You are a content writer.",
    model: "gpt-4",
  },
]);

// Execute task with council
const result = await rig.executeCouncil(councilId, "Write a research report on AI");
console.log(result);
// Output includes responses from all agents

Tools

// List all available tools
const tools = await rig.listTools();

// Search for relevant tools
const searchTools = await rig.searchTools("research", 5);
// Returns: web_search, academic_search, etc.

RAG (Retrieval-Augmented Generation)

// Add document to vector store
const docId = await rig.addDocument(
  "AI agents are transforming software development...",
  { source: "blog", author: "admin" }
);

// Search documents semantically
const results = await rig.searchDocuments("AI in software development", 5);

// Get specific document
const doc = await rig.getDocument(docId);

// Delete document
await rig.deleteDocument(docId);

HTTP API

Agents

Method Endpoint Description
POST /api/agents Create agent
GET /api/agents List agents
GET /api/agents/:id Get agent
POST /api/agents/:id/prompt Execute prompt
DELETE /api/agents/:id Delete agent

Councils

Method Endpoint Description
POST /api/councils Create council
GET /api/councils List councils
POST /api/councils/:id/execute Execute council task

Tools

Method Endpoint Description
GET /api/tools List all tools
POST /api/tools/search Search tools

Documents

Method Endpoint Description
POST /api/documents Add document
GET /api/documents List documents
POST /api/documents/search Search documents
GET /api/documents/:id Get document
DELETE /api/documents/:id Delete document

Configuration

Environment Variables

# Rig Service Configuration
RIG_HOST=127.0.0.1
RIG_PORT=8080
RIG_DATABASE_PATH=rig-store.db

# Model Providers
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
QWEN_API_KEY=...

# Defaults
RIG_DEFAULT_PROVIDER=openai
RIG_DEFAULT_MODEL=gpt-4

Rig Service Config

Edit rig-service/.env:

RIG_HOST=127.0.0.1
RIG_PORT=8080
RIG_DATABASE_PATH=./data/rig-store.db
OPENAI_API_KEY=your-key-here

Use Cases

1. Research Assistant

const researcher = await rig.createAgent({
  name: "researcher",
  preamble: "You are a research specialist. Find accurate, up-to-date information.",
  model: "gpt-4",
});

// Add research papers to vector store
await rig.addDocument("Paper: Attention Is All You Need...", { type: "paper" });
await rig.addDocument("Paper: BERT: Pre-training...", { type: "paper" });

// Search and execute
const context = await rig.searchDocuments("transformer architecture", 3);
const result = await rig.executePrompt(
  researcher,
  `Based on this context: ${context.map(d => d.content).join("\n")}, explain transformers.`
);

2. Code Review Council

const councilId = await rig.createCouncil("Code Review Team", [
  {
    name: "security",
    preamble: "You are a security expert. Review code for vulnerabilities.",
  },
  {
    name: "performance",
    preamble: "You are a performance expert. Identify bottlenecks.",
  },
  {
    name: "style",
    preamble: "You are a code style expert. Ensure clean, maintainable code.",
  },
]);

const review = await rig.executeCouncil(councilId, `
Review this code:
\`\`\`typescript
${code}
\`\`\`
`);

3. Content Creation Pipeline

// Create specialized agents
const researcher = await rig.createAgent({
  name: "researcher",
  preamble: "Research topics thoroughly.",
});

const writer = await rig.createAgent({
  name: "writer",
  preamble: "Write engaging content.",
});

const editor = await rig.createAgent({
  name: "editor",
  preamble: "Edit and polish content.",
});

// Or use a council
const councilId = await rig.createCouncil("Content Team", [
  { name: "researcher", preamble: "Research topics thoroughly." },
  { name: "writer", preamble: "Write engaging content." },
  { name: "editor", preamble: "Edit and polish content." },
]);

const article = await rig.executeCouncil(councilId, "Write an article about AI agents");

Building Rig Service

Prerequisites

  • Rust 1.70+
  • Cargo

Build

cd rig-service

# Debug build
cargo build

# Release build (optimized)
cargo build --release

# Run
cargo run

Cross-Platform

# Linux/macOS
cargo build --release

# Windows (MSVC)
cargo build --release

# Cross-compile for Linux from macOS
cargo install cross
cross build --release --target x86_64-unknown-linux-gnu

Troubleshooting

Rig Service Won't Start

# Check if port is in use
lsof -i :8080

# Check logs
RUST_LOG=debug cargo run

Connection Issues

// Test connection
const rig = initRigClient("127.0.0.1", 8080);
const healthy = await rig.health();
console.log("Rig healthy:", healthy);

Vector Store Issues

# Reset database
rm rig-store.db

# Check document count via API
curl http://127.0.0.1:8080/api/documents

Performance

Metric QwenClaw (TS) Rig (Rust)
Startup ~2-3s ~0.5s
Memory ~200MB ~50MB
Tool Lookup O(n) O(log n)
Concurrent Event loop Native threads

Next Steps

  1. Start Rig Service: cd rig-service && cargo run
  2. Initialize Client: initRigClient() in your code
  3. Create Agents: Define specialized agents for tasks
  4. Use Councils: Orchestrate multi-agent workflows
  5. Add RAG: Store and search documents semantically

Resources

License

MIT - Same as QwenClaw

Reference in New Issue View Git Blame Copy Permalink