# Povver MCP Server — Full Reference

> Connect AI assistants to your Povver training data. Povver's MCP server lets AI assistants read your training history, analyze progress, and answer questions about your workouts. Works with Claude Desktop and any MCP-compatible client.

## Overview
Povver is an AI strength training app for iOS. Its MCP (Model Context Protocol) server gives MCP-compatible AI clients secure, user-permissioned access to your training data: workout history, exercise progress, muscle-group analysis, training recommendations, injury memories, and routine/template management.

## Requirements
- Povver Premium subscription (every request is premium-gated; non-premium requests are rejected).
- Authenticate as the same account you use in the Povver iOS app.

## Compatible clients
- Claude Desktop — add as a custom connector (OAuth).
- Claude.ai — via the connector directory.
- Any MCP-compatible client — OAuth 2.1 (PKCE) or an API key.

## What you can do
- Pull your workout history into a conversation.
- See your Strength Climb — how much your lifts are climbing and which are leading.
- Ask about progress on specific exercises ("am I getting stronger on bench press?").
- Analyze your training volume by muscle group.
- Give an AI assistant context about your training when you ask for advice.
- Build custom analysis or exports on top of your own data.

## Authentication
Two methods, both sent as a bearer token to the MCP endpoint:
- OAuth 2.1 (recommended for Claude Desktop / Claude.ai): add the connector, sign in on the Povver consent page, approve. Access tokens are prefixed "pvt_" and refreshed automatically.
- API key (scripts and other clients): generate a key in the Povver iOS app under Settings → MCP Server. Keys are prefixed "pvk_" and shown once. Send as "Authorization: Bearer pvk_...".

## Setup — Claude Desktop / Claude.ai
1. Open Settings → Connectors → Add custom connector.
2. Enter the URL: https://mcp.povver.ai/mcp
3. Sign in on the Povver page with your Apple/Google/email account.
4. Approve access. Done.

## Setup — API key
1. Povver iOS app → Settings → MCP Server → generate an API key (pvk_...).
2. Configure your client with "Authorization: Bearer pvk_...".
3. Point the client at https://mcp.povver.ai/mcp

## Tools (32)
### Read
- check_connection: Verify the connection and return basic account info (name, subscription, whether you have an active routine and logged workouts).
- get_user_profile: The user's coaching profile: goal, experience level, target training days/week, equipment preference, height, weight (with units), timezone, member-since — to tailor advice to their background and constraints.
- get_training_snapshot: Compact overview of your training setup: profile, active routine, next workout, last 10 sessions, and strength records.
- list_routines: List all your routines with IDs, template IDs, frequency, and which one is active.
- get_routine: Get one routine with its template names and exercise summaries.
- list_templates: List all your workout templates (names and IDs).
- get_template: Get one template with its full exercise list and set prescriptions (reps, weight, RIR).
- list_workouts: List recent workouts as summaries with aggregate analytics (volume, sets, reps).
- get_workout: Get one workout with full set-level data (weight, reps, RIR, set type) and computed metrics.
- search_exercises: Search the exercise catalog by name or keyword; returns name, ID, muscle groups, and equipment.
- get_strength_climb: Your Strength Climb — the headline strength-progress signal at the top of the Intelligence tab: median % gain across qualifying lifts, per-state counts (climbing/holding/stalling/deloading/building), the leading lift, the climb line, and per-lift constituents. The PRIMARY strength signal — prefer it over the deprecated training score.
- get_training_insights: AI-generated insights and your latest weekly review: observations, guardrail alerts, fatigue, balance, trends, the strength_climb, training_context (volume completion, adherence, fatigue), and per-muscle muscle_volume (hard sets + zone). Also carries the deprecated training score (0-10) for now — prefer strength_climb + training_context.
- get_muscle_state: A muscle group's synthesized assessment: weekly hard sets, effective volume with volume zone + MEV/MAV/MRV targets, fatigue (ACWR), plateau status, e1RM trends, periodization phase, per-muscle strength_climb, and reasoning.
- get_muscle_group_progress: Raw weekly progress series for a muscle group (volume, set counts, e1RM) for charting and analysis.
- get_exercise_progress: Up to 8 weeks of progress for a specific lift: weekly e1RM trend, personal records, plateau detection, last session, the authoritative strength_state (progressing/holding/stalling) matching the iOS lift detail page, and strength_climb (indexed_pct + now-vs-baseline e1RM).
- query_sets: Query raw set-level data with flexible filters (exercise, muscle group, muscle, IDs) for custom analysis or export.
- get_training_status: Lightweight status: weekly adherence vs goal, next scheduled workout, last workout date, and days since training.

### Recommendations
- get_recommendations: Get pending training recommendations with rationale and confidence.
- review_recommendation: Accept, reject, or revert a recommendation — mutations are immediate.

### Memory
- list_memories: List your memories filtered by category (injury, preference, goal, personal) and status.
- get_memory: Get full details on a single memory, including status history and severity.
- get_recent_suppressions: List recommendations that were suppressed by injury memories, with the reasons.
- update_memory_status: Update an injury memory's lifecycle status (active, monitoring, resolved) with an audit note.
- save_injury_memory: Report a new injury or pain (body area, severity, affected exercises) so it influences training.

### Write — routines
- create_routine: Create a routine from existing template IDs (your first routine becomes active automatically).
- update_routine: Update a routine's name, frequency, or template order.
- set_active_routine: Set which routine is active (it determines your next workout).
- delete_routine: Permanently delete a routine (its templates are not deleted).

### Write — templates
- create_template: Create a workout template with exercises and set prescriptions.
- update_template: Update a template's name, description, or exercise list.
- delete_template: Permanently delete a template and remove it from any routines.

### Write — workouts
- delete_workout: Permanently delete a logged workout and its analytics data.

## Example prompts (44)
Plain-language requests; the assistant chains the tools above and reasons over the results.

### Use it as a coach (it chains tools for you)
- "Act as my strength coach: review my recent training, find my biggest weakness, and update my program to fix it."
- "Every Sunday, review my week and adjust next week’s plan."
- "Audit my whole program against my goals and flag anything that’s off."

### Track strength & spot plateaus
- "How’s my bench press trending over the last 8 weeks — am I still progressing or have I plateaued?"
- "Which of my main lifts are stalling, and which are still moving?"
- "What are my current estimated 1RMs across all my lifts?"
- "Did I set any personal records this month?"
- "Compare my squat and deadlift progress over the last 12 weeks."

### Understand volume & muscle balance
- "Break down my weekly training volume by muscle group."
- "Which muscle groups am I under-training, and where am I doing junk volume?"
- "Is my chest-to-back volume balanced? What about quads vs hamstrings?"
- "Show my chest effective volume and hard sets over the last 8 weeks."
- "Am I doing enough direct arm work?"

### Manage fatigue, recovery & deloads
- "Am I overreaching anywhere? Check my acute-to-chronic workload."
- "Do I need a deload this week, and for which muscle groups?"
- "Which muscles are carrying the most fatigue right now?"
- "Has my effort (RIR) been creeping up — am I grinding sets?"

### Plan your next session
- "What should I train today?"
- "Give me today’s workout with target weights, sets, and reps."
- "I’ve only got 45 minutes and dumbbells today — adapt my session."
- "What weight should I use on bench next session?"

### Build & adjust programs
- "Build me a 4-day upper/lower routine with the equipment I have."
- "My rear delts are lagging — add targeted volume to my current split."
- "Swap barbell back squat for a knee-friendly alternative across all my templates."
- "Turn my current routine into a 5-day push/pull/legs."
- "Design a 6-week arm specialization block."

### Act on Povver’s recommendations
- "What is Povver recommending for me right now, and why?"
- "Walk me through the rationale before I apply any progression changes."
- "Accept the recommendations you agree with and tell me which you skipped."

### Work around injuries & constraints
- "My left shoulder hurts on overhead pressing — log it and stop progressing those lifts."
- "Why didn’t my bench press get a progression last week?"
- "My knee feels better now — mark that injury resolved."
- "Rebuild this week’s leg day to avoid loaded knee flexion."

### Weekly & monthly reviews
- "Review my training week: volume, adherence, fatigue, and what to change."
- "How consistent have I been over the last 3 months — sessions vs my goal?"
- "Summarize how my training has progressed since I started."
- "When did I last train each muscle group?"

### Export & custom analysis
- "Export every squat set I’ve logged this year as a table."
- "What’s my total training volume (tonnage) this month?"
- "Pull all my chest sets above 80% of my estimated 1RM."
- "Build a CSV of my top set per workout for deadlift."

### Explore the exercise library
- "Find cable exercises for rear delts."
- "What are good hamstring exercises I can do with just dumbbells?"
- "List alternatives to barbell bench press for a tweaked shoulder."

## Endpoints
- MCP protocol: https://mcp.povver.ai/mcp (bearer auth; POST JSON-RPC, SSE supported)
- OAuth discovery: https://mcp.povver.ai/.well-known/oauth-authorization-server
- OAuth flow: https://mcp.povver.ai/authorize, /token, /register
- Health: https://mcp.povver.ai/health
- Docs (HTML): https://mcp.povver.ai/
- This file: https://mcp.povver.ai/llms-full.txt

## Notes
- The MCP endpoint is /mcp. The root path also accepts authenticated MCP requests for backward compatibility, but new integrations should use /mcp.
- Write tools (create/update/delete routines and templates, delete workouts, save/update injury memories, review recommendations) change your data; the iOS app remains the canonical place to log workouts.

## Links
- Website: https://povver.ai
- App Store: https://apps.apple.com/app/id6759248585
- Privacy: https://povver.ai/privacy
- Terms: https://povver.ai/tos