Redmine

What the Redmine MCP server does

The Redmine MCP server gives MCP-compatible AI hosts access to Redmine through the redmine-cli project. It exposes the command-line client’s Redmine operations as typed tool calls, allowing an agent to work with issues, projects, time entries, wiki pages, and other supported areas described by the CLI.

Operations are read-only by default. This makes the server suitable for agents that need to look up project information, inspect issues, or retrieve time and wiki data without automatically changing the Redmine instance. Write access can be controlled with per-group and per-tool allow- and deny-lists when write operations are needed.

How it works

Start the built-in MCP server with redmine mcp serve, then register that process with an MCP host. The server uses the Redmine connection and authentication configured for the CLI and presents the available CLI operations to the host as typed calls rather than requiring the agent to construct command-line invocations.

The underlying project supports Redmine versions 4.x, 5.x, 6.x, and 7.x. The repository also documents a Docker-based MCP option using the ghcr.io/redmine-mcp image. A Docker launch passes the Redmine server address and API key into the container through environment variables.

Setup and configuration

Install the CLI through Homebrew, the quick install script, Go, or a platform-specific release archive. Homebrew supports macOS and Linux; release archives are listed for Linux, macOS, and Windows architectures. After installation, configure the Redmine instance and API key with redmine auth login.

To run the Redmine MCP server locally, use:

redmine mcp serve

For Docker, the documented command supplies REDMINE_SERVER and REDMINE_API_KEY to the container:

docker run -i --rm -e REDMINE_SERVER -e REDMINE_API_KEY ghcr.io/redmine-mcp

Host-specific configuration guidance is documented for Claude Desktop, Cursor, Zed, and VS Code. The project also provides installation paths for Claude Code, Codex CLI, and Gemini CLI.

Tools and capabilities

The MCP interface exposes the same operations as redmine-cli, including capabilities for:

• Reading and working with Redmine issues
• Accessing project information
• Recording or inspecting time entries
• Reading and working with wiki pages
• Using typed MCP calls instead of raw CLI commands
• Restricting access by tool group or individual tool

The exact command and tool set can be inspected through the CLI’s help output and the project’s AI agent integration documentation.

Limitations and notes

The Redmine MCP server is read-only by default, so agents should not be expected to perform changes unless write tools are explicitly enabled through the available gating configuration. The provided material does not enumerate every MCP tool or describe the exact allow-list and deny-list syntax.

A reachable Redmine instance and valid authentication are required. Compatibility with Redmine 4.x through 7.x is indicated in the project materials, while local end-to-end testing is documented for versions 4.2, 5.1, 6.1, and 7.0. The MCP server is part of the redmine-cli project, so installation and updates follow that project’s release process.