docker/compose-for-agents

Build and run AI agents using Docker Compose. A collection of ready-to-use examples for orchestrating open-source LLMs, tools, and agent runtimes.

What it solves

This repository provides a collection of self-contained demos that simplify the deployment and orchestration of AI agents. It allows developers to quickly spin up complex agent systems—including those using multi-agent frameworks and Model Context Protocol (MCP) servers—using Docker Compose, reducing the setup friction for local or cloud environments.

How it works

Each demo is configured as a Docker Compose project. Users can run these agents either locally using the Docker Model Runner (utilizing a GPU) or via remote APIs like OpenAI. The demos integrate various agent frameworks (such as CrewAI, LangGraph, and Agno) and connect them to external tools and data sources via MCPs (like DuckDuckGo, GitHub, and Wikipedia).

Who it’s for

Developers and AI engineers who want to explore, test, and and deploy agentic workflows and multi-agent systems without manually configuring the underlying infrastructure, models, and tool integrations.

Highlights

  • Multi-Framework Support: Includes examples for CrewAI, LangGraph, Agno, ADK, and Vercel AI-SDK.
  • Flexible Model Execution: Supports both local model execution via Docker Model Runner and remote API-based models (e.g., OpenAI).
  • MCP Integration: Demonstrates the use of Model Context Protocol (MCP) servers for real-world capabilities like web search, database access, and email.
  • Rapid Deployment: Uses docker compose up --build to automate the environment setup for various agentic use cases.

Related

  • Project
    atlassian/atlassian-mcp-server

    A cloud‑hosted bridge that lets AI tools (ChatGPT, Claude, VS Code Copilot, Cursor, etc.) securely read, search, and write Atlassian Cloud data (Jira, Confluence, Bitbucket, Loom, Compass, and the Teamwork Graph) via the open Model Context Protocol. Install the connector, authenticate with OAuth 2.1 or an API token, and issue natural‑language commands that the server translates into Atlassian API calls.

  • Project
    rl-tools/rl-tools

    A fast, portable C++ library for deep reinforcement learning and continuous control, optimized for everything from microcontrollers to GPUs.

  • Project
    google/adk-web

    A developer UI for the Google Agent Development Kit that simplifies the building, debugging, and orchestration of AI agents.

  • Project
    kunal12203/GrapeRoot

    A context engine for AI coding assistants that builds a semantic graph of a codebase to pre-load relevant code into prompts, reducing token usage and response times.