Building AI agents in the real world often involves more than just making model calls \u2014 it requires integrating with external tools, handling complex workflows, and ensuring the solution can scale in production.<\/p>\n
In this post, we\u2019ll walk through a real-world developer setup for creating an agent using the Docker MCP Toolkit.<\/p>\n
To make things concrete, I\u2019ve built an agent that takes a Git repository as input and can answer questions about its contents \u2014 whether it\u2019s explaining the purpose of a function, summarizing a module, or finding where a specific API call is made. This simple but practical use case serves as a foundation for exploring how agents can interact with real-world data sources and respond intelligently.<\/p>\n
I built and ran it using the Docker MCP Toolkit, which made setup and integration fast, portable, and repeatable. This blog walks you through that developer setup and explains why Docker MCP is a game changer for building and running agents.<\/p>\n
The goal: Build an AI agent that can connect to a GitHub repository, retrieve relevant code or metadata, and answer developer questions in plain language.<\/p>\n
Example queries:<\/p>\n
\u201cSummarize this repo: https:\/\/github.com\/owner\/repo\u201d<\/p>\n
\u201cWhere is the authentication logic implemented?\u201d<\/p>\n
\u201cList main modules and their purpose.\u201d<\/p>\n
\u201cExplain the function parse_config and show where it\u2019s used.\u201d<\/p>\n
This goes beyond a simple code demo \u2014 it reflects how developers work in real-world environments<\/strong> The agent acts like a code-aware teammate you can query anytime.<\/p>\n The MCP Gateway handles tooling integration (GitHub API) without bloating the agent code.<\/p>\n Docker Compose ties the environment together so it runs the same in dev, staging, or production.<\/p>\n Without MCP Toolkit, you\u2019d spend hours wiring up API SDKs, managing auth tokens, and troubleshooting environment differences.<\/p>\n With MCP Toolkit:<\/p>\n Containerized connectors \u2013 Run the GitHub MCP Gateway as a ready-made service (docker\/mcp-gateway:latest), no SDK setup required.<\/p>\n Consistent environments \u2013 The container image has fixed dependencies, so the setup works identically for every team member.<\/p>\n Rapid integration \u2013 The agent connects to the gateway over HTTP; adding a new tool is as simple as adding a new container.<\/p>\n Iterate faster \u2013 Restart or swap services in seconds using docker compose.<\/p>\n Focus on logic, not plumbing \u2013 The gateway handles the GitHub-specific heavy lifting while you focus on prompt design, reasoning, and multi-agent orchestration.<\/p>\n Running everything via Docker Compose means you treat the entire agent environment as a single deployable unit:<\/p>\n One-command startup \u2013 docker compose up brings up the MCP Gateway (and your agent, if containerized) together.<\/p>\n Service orchestration \u2013 Compose ensures dependencies start in the right order.<\/p>\n Internal networking \u2013 Services talk to each other by name (http:\/\/mcp-gateway-github:8080) without manual port wrangling.<\/p>\n Scaling \u2013 Run multiple agent instances for concurrent requests.<\/p>\n Unified logging \u2013 View all logs in one place for easier debugging.<\/p>\n This setup connects a developer\u2019s local agent to GitHub through a Dockerized MCP Gateway, with Docker Compose orchestrating the environment. Here\u2019s how it works step-by-step:<\/p>\n User Interaction<\/p>\n The developer runs the agent from a CLI or terminal.<\/p>\n They type a question about a GitHub repository \u2014 e.g., \u201cWhere is the authentication logic implemented?\u201d<\/p>\n Agent Processing<\/p>\n The Agent (LLM + MCPTools) receives the question.<\/p>\n The agent determines that it needs repository data and issues a tool call via MCPTools.<\/p>\n MCPTools \u2192 MCP Gateway<\/p>\n \u00a0MCPTools sends the request using streamable-http to the MCP Gateway running in Docker.<\/p>\n This gateway is defined in docker-compose.yml and configured for the GitHub server (–servers=github –port=8080).<\/p>\n GitHub Integration<\/p>\n The MCP Gateway handles all GitHub API interactions \u2014 listing files, retrieving content, searching code \u2014 and returns structured results to the agent.<\/p>\n LLM Reasoning<\/p>\n The agent sends the retrieved GitHub context to OpenAI GPT-4o as part of a prompt.<\/p>\n \u00a0The LLM reasons over the data and generates a clear, context-rich answer.<\/p>\n Response to User<\/p>\n The agent prints the final answer back to the CLI, often with file names and line references.<\/p>\n
<\/strong><\/p>\nRole of Docker MCP Toolkit<\/h2>\n
Role of Docker Compose\u00a0<\/h2>\n
Architecture Overview<\/h2>\n
Code Reference & File Roles<\/h2>\n