CodeAct
Implementation of the CodeAct agentic framework with Docker containers for security, MCP servers for tool integrations, and multi-agent support.
What is CodeAct?
CodeAct is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to implementation of the codeact agentic framework with docker containers for security, mcp servers for tool integrations, and multi-agent support.
Implementation of the CodeAct agentic framework with Docker containers for security, MCP servers for tool integrations, and multi-agent support.
This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- Implementation of the CodeAct agentic framework with Docker
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx codearktConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use CodeAct
CodeArkt is a Python implementation of the CodeAct agentic framework that lets you build and run autonomous coding agents capable of writing, executing, and iterating on Python code inside a secure Docker sandbox. It provides first-class MCP tool integration for auto-discovering and registering external tool servers, supports hierarchical multi-agent orchestration (manager/worker patterns), and ships with both a Gradio web UI and a terminal client for interacting with agents. It is designed for teams that want production-ready autonomous agents with typed code, CI support, and an Apache-2.0 license.
Prerequisites
- Python 3.12 or higher installed
- Docker installed and running (required for the secure sandbox execution environment)
- An LLM provider API key (e.g., OpenRouter, OpenAI-compatible endpoint)
- uv package manager recommended (pip also works)
- An MCP-compatible client or the built-in Gradio/terminal UI
Install CodeArkt
Install the codearkt package from PyPI. Python 3.12 or higher is required.
pip install codearktSet your LLM provider API key
CodeArkt uses OpenRouter or any OpenAI-compatible provider. Export your API key as an environment variable.
export OPENROUTER_API_KEY="your_openrouter_api_key_here"Start any MCP tool servers you want the agent to use
CodeArkt discovers tools from external MCP servers. Start them before launching your agent. For example, start an academia MCP server on port 5056.
python -m academia_mcp --port 5056Create and run a CodeActAgent server
Define an agent with a name, description, LLM config, and the tool names it should have access to. Then start the server which exposes an MCP proxy and agentic HTTP endpoints.
import os
from codearkt.codeact import CodeActAgent
from codearkt.llm import LLM
from codearkt.server import run_server
mcp_config = {
"mcpServers": {
"academia": {"url": "http://0.0.0.0:5056/mcp", "transport": "streamable-http"}
}
}
agent = CodeActAgent(
name="manager",
description="A simple coding agent",
llm=LLM(
model_name="deepseek/deepseek-chat-v3-0324",
api_key=os.getenv("OPENROUTER_API_KEY")
),
tool_names=["arxiv_download", "arxiv_search"],
)
run_server(agent, mcp_config, port=5055)Interact with the agent
Use the terminal client or Gradio UI to send tasks to the running agent. The agent will reason, write code, execute it in Docker, and iterate.
# Terminal client
uv run -m codearkt.terminal --port 5055
# Or Gradio web UI
uv run -m codearkt.gradio --port 5055CodeAct Examples
Client configuration
CodeArkt runs as a standalone server that your client connects to via HTTP. The agent server itself consumes MCP servers as tools.
{
"mcpServers": {
"codearkt-agent": {
"url": "http://localhost:5055/mcp",
"transport": "streamable-http"
}
}
}Prompts to try
Once the agent server is running and connected, send tasks that require writing and executing code.
- "Find and summarize the abstract of arxiv paper 2402.01030"
- "Write a Python script to compute the first 20 Fibonacci numbers and run it"
- "Download the top 5 papers on RAG from arxiv this month and create a summary table"
- "Implement a binary search function, write tests for it, and execute them"Troubleshooting CodeAct
Docker sandbox fails to execute code
Ensure Docker is running and the current user has permission to run containers (`docker ps` should work). CodeArkt uses ephemeral containers, so the Docker daemon must be accessible.
Agent cannot find MCP tools
Make sure your external MCP servers are started and reachable at the URLs specified in the `mcp_config` dict before calling `run_server`. Verify with `curl http://0.0.0.0:5056/mcp` or equivalent.
ImportError or Python version mismatch
CodeArkt requires Python 3.12 or higher. Check your version with `python --version` and use a version manager like pyenv or uv to switch to 3.12+.
Frequently Asked Questions about CodeAct
What is CodeAct?
CodeAct is a Model Context Protocol (MCP) server that implementation of the codeact agentic framework with docker containers for security, mcp servers for tool integrations, and multi-agent support. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install CodeAct?
Follow the installation instructions on the CodeAct GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with CodeAct?
CodeAct works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is CodeAct free to use?
Yes, CodeAct is open source and available under the Apache-2.0 license. You can use it freely in both personal and commercial projects.
CodeAct Alternatives — Similar Coding Agents Servers
Looking for alternatives to CodeAct? Here are other popular coding agents servers you can use with Claude, Cursor, and VS Code.
Dify
★ 142.2kProduction-ready platform for agentic workflow development.
Ruflo
★ 54.0k🌊 The leading agent orchestration platform for Claude. Deploy intelligent multi-agent swarms, coordinate autonomous workflows, and build conversational AI systems. Features enterprise-grade architecture, self-learning swarm intelligence, RAG integrat
Goose
★ 45.7kan open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM
Antigravity Awesome Skills
★ 38.3kInstallable GitHub library of 1,400+ agentic skills for Claude Code, Cursor, Codex CLI, Gemini CLI, Antigravity, and more. Includes installer CLI, bundles, workflows, and official/community skill collections.
AgentScope
★ 25.5kBuild and run agents you can see, understand and trust.
Serena
★ 24.5kA coding agent toolkit that provides IDE-like semantic code retrieval and editing tools, enabling LLMs to efficiently navigate and modify codebases using symbol-level operations instead of basic file reading and string replacements.
Browse More Coding Agents MCP Servers
Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up CodeAct in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use CodeAct?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.