# Evertrain

> Evertrain is an AI-powered personal training platform. It gives AI assistants direct read/write access to a user's training data — programs, workouts, calendar, and coaching facts — via the Model Context Protocol (MCP).

## What Evertrain does

Evertrain connects AI assistants (Claude, ChatGPT, and any MCP-compatible agent) to a user's personal training database. The AI can read the user's full exercise history, scheduled program, and coaching facts (preferences, restrictions, goals), and can write — creating programs, scheduling workouts, logging activity, and proposing coaching fact updates — all without the user switching apps.

The mobile app (iOS and Android) is where users train: they follow workouts, log sets, track rest timers, and review AI-authored coaching notes. The web console (my.evertrain.app) is where they manage programs, connect AI agents, and review their data. The AI acts through the MCP layer; Evertrain is the data and the connection.

## Core capabilities (available to connected AI agents)

- `get_user_profile` — retrieve the user's profile, units, and training preferences
- `get_coaching_facts` — retrieve structured coaching facts: goals, restrictions, equipment, history
- `get_training_history` — full workout log with sets, reps, weights, and notes
- `get_scheduled_calendar` — upcoming scheduled workouts and program weeks
- `get_user_programs` — list of all training programs (active, archived)
- `get_program` / `get_workout` — full detail on a specific program or workout
- `create_program` — create a complete multi-week training program
- `create_workout` — create a single workout session
- `update_program` / `update_future_workout` — edit an existing program or upcoming workout
- `schedule_program` — schedule a program to start on a specific date
- `reschedule_workout` — move a scheduled workout to a different date
- `append_week` / `insert_week` — extend a program or insert a deload/recovery week
- `shift_from` — shift all workouts from a given date forward
- `propose_preference` / `propose_restriction` / `propose_current_phase` — propose coaching fact updates for user approval
- `post_activity` — log a completed workout or activity
- `get_daily_notes` — retrieve AI-authored coaching notes for specific dates
- `get_exercise_library` — browse the full exercise library with search and filtering
- `archive_program` / `delete_program` — manage program lifecycle
- `submit_feature_request` — submit a feature request to the Evertrain team

## Pricing

- **Free** — access to the app and web console; import and manage programs; no AI agent connection.
- **Evertrain Connect** — paid subscription; connects the user's AI assistant via MCP; unlocks all agent capabilities listed above.

## Technical integration

Evertrain exposes an MCP server at `mcp.evertrain.app`. AI agents connect via OAuth. Each user's agent session is scoped to that user's data only. The MCP server is built on Firebase Cloud Functions (Node.js 20) with Genkit.

## Developer and agent documentation

- Agent capabilities and usage guide: available to connected agents via the MCP server
- Feature requests: `submit_feature_request` tool (available to connected agents)
- Web console: https://my.evertrain.app
- MCP endpoint: https://mcp.evertrain.app

## Who built this

Evertrain is an independent product. The focus is on giving AI assistants the data and tools to be genuinely useful personal training coaches — not replacing the AI, but connecting it to the user's real training life.

## Contact

For questions, support, or partnership inquiries: contact via the web console at https://my.evertrain.app.