wger-agent¶
Wger Workout Manager MCP server + A2A agent for the Arbora ecosystem — exercise database, workout routines, nutrition plans, body measurements, and progress tracking, exposed as typed, deterministic tools.
Official documentation
This site is the canonical reference for wger-agent, maintained alongside every
release.
Overview¶
wger-agent wraps the Wger Workout Manager REST API with
action-routed MCP tools and a Pydantic-AI graph agent. It provides:
WgerApi— a unified Python client composed of domain-specific sub-clients (routine, exercise, nutrition, workout, body, user) over the Wger REST surface.- Action-routed MCP tools across seven togglable domains, registered through the
scion-sdkFastMCP middleware to minimize token overhead in LLM contexts. - An integrated graph agent (the
wger-agentconsole script) that speaks the Agent Control Protocol and the Agent Web UI, with a confidence-gated router that enables only the tools relevant to each request.
Explore the documentation¶
- :material-rocket-start: Installation — pip, source, extras, and the prebuilt Docker image.
- Deployment — run the MCP and agent servers, Docker Compose, Caddy + Technitium.
- Usage — the MCP tools, the
WgerApiclient, and the agent CLI. - Backing Platform — deploy the Wger Workout Manager with Docker.
- Architecture — the standardized agent-package pattern.
- Concepts — the
CONCEPT:WGER-*registry.
Quick start¶
Connect it to a Wger instance:
export WGER_URL=https://your-wger:8000
export WGER_API_KEY=your_api_key
wger-mcp --transport streamable-http --host 0.0.0.0 --port 8000
See Installation and Deployment for the full matrix (PyPI extras, Docker image, all transports, the agent server, reverse proxy, DNS).
Spec delivery¶
The spec delivery dashboard reports the generated delivery status for this connector.