Skip to content

Usage — MCP / API / CLI

wger-agent exposes the same capability three ways: as MCP tools an agent calls, as a Python API (WgerApi) you import, and as a graph agent CLI. The full tool surface is summarized in Overview.

As an MCP server

Once deployed, the server registers seven action-routed tool domains, each grouping related methods to keep the LLM context lean. Every domain is togglable through its environment variable (all default True):

Tool domain Toggle Representative actions
Routine ROUTINETOOL create_routine, get_routines, create_day, get_templates
Routine config ROUTINECONFIGTOOL create_sets_config, create_weight_config, get_repetitions_configs
Exercise EXERCISETOOL search_exercises, get_exercises, get_muscles, get_equipment
Workout WORKOUTTOOL create_workout_session, create_workout_log, get_workout_logs
Nutrition NUTRITIONTOOL create_nutrition_plan, log_nutrition, get_ingredients
Body BODYTOOL log_body_weight, log_measurement, get_weight_entries
User USERTOOL get_user_profile, get_user_statistics, get_languages

Example agent prompts that map onto these tools:

  • "Find a barbell bench-press exercise" → search_exercises
  • "Log a body weight of 80 kg for today" → log_body_weight
  • "Show my active workout routines" → get_routines

As a Python API

WgerApi is a unified client composed of domain-specific sub-clients, built from the environment via get_client():

from wger_agent.auth import get_client

api = get_client()        # reads WGER_URL / WGER_API_KEY from the environment / .env

# Reads
exercises = api.search_exercises("bench press")
muscles = api.get_muscles()
routines = api.get_routines()
profile = api.get_user_profile()

Build a client directly:

from wger_agent.api_client import WgerApi

api = WgerApi(
    base_url="https://your-wger:8000",
    token="your_api_key",
    verify=True,
)

ingredients = api.get_ingredients()
weight_entries = api.get_weight_entries()

Writes

The same client performs writes once it is configured with a valid token:

api.log_body_weight(weight=80.0)
api.create_nutrition_plan(description="Cutting plan")
api.create_workout_session(notes="Upper body")

As an agent CLI

The integrated Pydantic-AI graph agent runs from the wger-agent console script. A confidence-gated router classifies each request and enables only the relevant tool domain, then a domain node executes against the MCP server.

export WGER_URL=https://your-wger:8000
export WGER_API_KEY=your_api_key
export MCP_URL=http://localhost:8000/mcp

wger-agent --provider openai --model-id gpt-4o

Enable the Agent Web UI and OpenTelemetry tracing with --web and --otel. See Deployment for the full Compose recipe and port wiring.