The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Redmine listing page.
A command-line interface for Redmine project management.
Installation · Getting Started · AI Agents
This also installs shell completions for bash, zsh, and fish.
Auto-detects your OS and architecture, downloads the latest release with checksum verification, and installs to ~/.local/bin.
Grab the latest release for your platform from GitHub Releases:
| Platform | Architecture | Download |
|---|---|---|
| Linux | x86_64 | redmine-cli-linux-amd64.tar.gz |
| Linux | ARM64 | redmine-cli-linux-arm64.tar.gz |
| macOS | Intel | redmine-cli-darwin-amd64.tar.gz |
| macOS | Apple Silicon | redmine-cli-darwin-arm64.tar.gz |
| Windows | x86_64 | redmine-cli-windows-amd64.zip |
Downloads and verifies the latest release via SHA256 checksum before replacing the binary.
Run redmine --help to see all available commands.
-o json prints machine-readable JSON, and -q/--jq <expression> filters it (strings are printed raw, everything else as JSON). Errors always go to stderr; in JSON mode they are a single-line {"error": {...}} envelope on the last line of stderr, stdout stays empty and the exit code is non-zero. See Output formats.
redmine-cli ships an Agent Skill (the open SKILL.md standard, supported by 35+ agents) and a built-in MCP server. Both are vendor-neutral - pick the one-liner that matches your tool:
| Tool | One-line install |
|---|---|
| Any agent (skill) | npx skills add aarondpn/redmine-cli |
| Any MCP host | redmine mcp serve (then add to host config) |
| Docker (MCP only) | docker run -i --rm -e REDMINE_SERVER -e REDMINE_API_KEY ghcr.io/aarondpn/redmine-mcp |
| Claude Code | /plugin marketplace add aarondpn/redmine-cli then /plugin install redmine |
| Codex CLI | codex plugin marketplace add aarondpn/redmine-cli then /plugins (install Redmine from the added marketplace) |
| Gemini CLI | gemini extensions install https://github.com/aarondpn/redmine-cli |
The skill teaches the agent the non-obvious parts of the CLI (output formats, name resolution, pagination, common workflows). The MCP server exposes the same operations as typed tool calls, read-only by default, with per-group / per-tool allow- and deny-lists.
Full configuration, host-specific snippets (Claude Desktop, Cursor, Zed, VS Code), and write-tool gating: see the AI Agent Integration guide.
Local E2E testing against a real Redmine instance (Docker, supported on Redmine 4.2, 5.1, 6.1, 7.0): see e2e/README.md.