myxagent 0.2.35

Multi-Modal AI Agent System

xAgent

A production-ready AI Agent framework focused on easy start and scalable deployment.

• ✅ Chat via CLI / Python API / HTTP Server
• ✅ Built-in Web UI and streaming responses
• ✅ Tool calling, MCP integration, image input
• ✅ Multi-user + multi-session support
• ✅ Memory and multi-agent workflows

3-Minute Quick Start

1) Install

pip install myxagent

2) Set environment variable

export OPENAI_API_KEY=your_openai_api_key

3) Start using xAgent

# Interactive CLI
xagent-cli

# Or ask one question
xagent-cli --ask "Hello"

Configure with agent.yaml

If you want to customize the agent prompt, model, tools, or server port, create a YAML config file.

1) Generate a starter config

xagent-cli --init

This creates:

• config/agent.yaml
• my_toolkit/ for custom tools

2) Edit config/agent.yaml

agent:
    name: "Assistant"
    system_prompt: |
        You are a helpful AI assistant.
        Answer clearly and accurately.
    model: "gpt-5-mini"

    capabilities:
        tools:
            - "web_search"

    storage_mode: "local"

server:
    host: "0.0.0.0"
    port: 8010

3) Run with your config

# CLI
xagent-cli --config config/agent.yaml --toolkit_path my_toolkit

# HTTP Server + Web UI
xagent-server --config config/agent.yaml --toolkit_path my_toolkit --open

If you do not use custom tools, you can omit --toolkit_path.

For more YAML options, see docs/configuration_reference.md.

Cloud Mode

Use cloud mode when you want shared/distributed conversation history and cloud memory.

1) Install cloud dependencies

pip install "myxagent[cloud]"

2) Set storage_mode: "cloud"

agent:
    storage_mode: "cloud"

3) Set required environment variables

export REDIS_URL=redis://localhost:6379/0
export UPSTASH_VECTOR_REST_URL=https://your-database.upstash.io
export UPSTASH_VECTOR_REST_TOKEN=your_token_here

For config-driven cloud mode, all three are required. If you want local-only execution, keep storage_mode: "local" and install myxagent only.

Most Common Usage

CLI

xagent-cli
xagent-cli --ask "What is the weather in Hangzhou?"

HTTP Server (API + Web UI)

xagent-server
# http://localhost:8010

# Open Web UI automatically
xagent-server --open

# API call example
curl -X POST "http://localhost:8010/chat" \
  -H "Content-Type: application/json" \
  -d '{
    "user_id": "user123",
    "session_id": "session456",
    "user_message": "Hello"
  }'

# Continuous conversation: keep same user_id + session_id
# Turn 1
curl -X POST "http://localhost:8010/chat" \
    -H "Content-Type: application/json" \
    -d '{
        "user_id": "alice",
        "session_id": "daily_chat",
        "user_message": "Remember that my favorite city is Hangzhou."
    }'

# Turn 2 (same session)
curl -X POST "http://localhost:8010/chat" \
    -H "Content-Type: application/json" \
    -d '{
        "user_id": "alice",
        "session_id": "daily_chat",
        "user_message": "What is my favorite city?"
    }'

# Image input via image_source (single image)
curl -X POST "http://localhost:8010/chat" \
    -H "Content-Type: application/json" \
    -d '{
        "user_id": "user123",
        "session_id": "image_session",
        "user_message": "Describe this image.",
        "image_source": "https://example.com/image.jpg"
    }'

# Image input via image_source (multiple images)
curl -X POST "http://localhost:8010/chat" \
    -H "Content-Type: application/json" \
    -d '{
        "user_id": "user123",
        "session_id": "image_session",
        "user_message": "Compare these images.",
        "image_source": [
            "https://example.com/image1.jpg",
            "https://example.com/image2.jpg"
        ]
    }'

# Image URL directly in message text (no image_source needed)
curl -X POST "http://localhost:8010/chat" \
    -H "Content-Type: application/json" \
    -d '{
        "user_id": "user123",
        "session_id": "image_in_message",
        "user_message": "What do you see in this image? https://example.com/cat.jpg"
    }'

Python API

import asyncio
from xagent.core import Agent

async def main():
    agent = Agent(model="gpt-5-mini")
    response = await agent.chat(
        user_message="Hello",
        user_id="user123",
        session_id="session456"
    )
    print(response)

asyncio.run(main())

Continuous Conversation (same session)

Use the same user_id and session_id to keep context across turns:

import asyncio
from xagent.core import Agent

async def main():
    agent = Agent(model="gpt-5-mini")

    user_id = "alice"
    session_id = "daily_chat"

    reply1 = await agent.chat(
        user_message="Remember that my favorite city is Hangzhou.",
        user_id=user_id,
        session_id=session_id,
    )
    print("Turn 1:", reply1)

    reply2 = await agent.chat(
        user_message="What is my favorite city?",
        user_id=user_id,
        session_id=session_id,
    )
    print("Turn 2:", reply2)

asyncio.run(main())

Image Input Support

image_source supports a single value or list, and each item can be an image URL, local file path, or base64 data URI.

import asyncio
from xagent.core import Agent

async def main():
    agent = Agent(model="gpt-5-mini")

    # Single image URL
    reply1 = await agent.chat(
        user_message="What do you see in this image?",
        user_id="user123",
        session_id="image_demo",
        image_source="https://example.com/image.jpg",
    )
    print("Single image:", reply1)

    # Multiple images (URL + local path)
    reply2 = await agent.chat(
        user_message="Compare these two images.",
        user_id="user123",
        session_id="image_demo",
        image_source=[
            "https://example.com/image1.jpg",
            "./local_image.png",
        ],
    )
    print("Multi-image:", reply2)

asyncio.run(main())

Recommended Learning Path

1. Quickly run it: this README
2. Project setup + config: xagent-cli --init
3. Pick your interface: CLI / HTTP / Python API
4. Then add advanced capabilities: memory, workflows, custom tools

Documentation Center

All technical details are now organized in docs/:

Start Here

• Documentation Index
• Getting Started
• Best Practices

Core References

• Configuration Reference
• API Reference
• Memory System
• Multi-Agent Workflows
• Workflow DSL

Advanced / Deployment

• HTTP Server Agent Passing
• Message Storage Inheritance
• Redis Cluster Support
• Docker Deployment

Examples

• Examples Overview
• Demo Examples
• Config Examples
• Toolkit Examples

Contributing

Contributions are welcome. Please open an issue or pull request on GitHub.

License

This project is licensed under the MIT License. See LICENSE.