Agent Trace CLI

A command-line tool for monitoring LangChain agents in real-time with the AXON dashboard.

Installation

Global Installation (Recommended)

npm install -g @axon-ai/cli

Local Installation

npm install @axon-ai/cli
npx axon-ai --help

Quick Start

1. Initialize Axon in your project:

   axon-ai init --project my-ai-project

2. Start the dashboard:

   axon-ai start

3. Add tracing to your LangChain agents:

   import { createTracer } from '@axon-ai/langchain-tracer';
      
   const tracer = createTracer({
     projectName: 'my-ai-project'
   });
      
   const model = new ChatOpenAI({
     modelName: 'gpt-3.5-turbo',
     callbacks: [tracer] // Add the tracer
   });

4. Run your agents and watch them in real-time!

Commands

axon-ai init

Initialize AXON in your current project.

axon-ai init [options]

Options:

• --project <name> - Project name (default: "default")
• --auto-start - Automatically start dashboard after initialization

Example:

axon-ai init --project my-ai-app --auto-start

axon-ai start

Start the AXON dashboard and enable tracing.

axon-ai start [options]

Options:

• -p, --port <port> - Backend server port (default: 3000)
• -d, --dashboard-port <port> - Dashboard port (default: 5173)
• --no-open - Don't automatically open dashboard in browser
• --project <name> - Project name for organizing traces

Example:

axon-ai start --port 3001 --dashboard-port 5174

agent-trace status

Check the status of AXON services.

axon-ai status

Shows:

• Project information
• Backend server status
• Dashboard status
• Quick action suggestions

axon-ai stop

Stop all AXON services.

axon-ai stop

axon-ai version

Show version information.

axon-ai version

Integration with LangChain

Basic Integration

import { createTracer } from '@axon-ai/langchain-tracer';
import { ChatOpenAI } from '@langchain/openai';

// Create tracer
const tracer = createTracer({
  projectName: 'my-project',
  endpoint: 'http://localhost:3000'
});

// Add to your model
const model = new ChatOpenAI({
  modelName: 'gpt-3.5-turbo',
  callbacks: [tracer]
});

Agent Integration

import { AgentExecutor, createOpenAIFunctionsAgent } from 'langchain/agents';

const agent = await createOpenAIFunctionsAgent({
  llm: model,
  tools: [searchTool, calculatorTool],
  prompt: agentPrompt
});

const agentExecutor = new AgentExecutor({
  agent,
  tools: [searchTool, calculatorTool],
  callbacks: [tracer] // Add tracer to executor too
});

Chain Integration

import { LLMChain } from 'langchain/chains';

const chain = new LLMChain({
  llm: model,
  prompt: myPrompt,
  callbacks: [tracer]
});

Configuration

After running axon-ai init, a .axon-ai/config.json file is created:

{
  "project": "my-project",
  "version": "1.0.0",
  "initialized": "2024-01-15T10:30:00.000Z",
  "backend": {
    "port": 3000,
    "host": "localhost"
  },
  "dashboard": {
    "port": 5173,
    "host": "localhost"
  }
}

Troubleshooting

Port Already in Use

If you get a "port already in use" error:

# Check what's using the port
lsof -i :3000

# Kill the process
kill -9 <PID>

# Or use different ports
axon-ai start --port 3001 --dashboard-port 5174

Services Not Starting

1. Check if ports are available:

   axon-ai status

2. Stop all services and restart:

   axon-ai stop
   axon-ai start

3. Check logs in the terminal where you started the services

Dashboard Not Opening

If the dashboard doesn't open automatically:

1. Check the status: axon-ai status
2. Manually open: http://localhost:5173 (or your configured port)
3. Make sure the backend is running on the correct port

Development

Building from Source

git clone https://github.com/yourusername/langchain-tracer/Axon.git
cd axon-ai
npm install
npm run build:cli

Running in Development

cd packages/cli
npm run dev

License

MIT License - see LICENSE for details.