Fitness

Auto-learns your fitness patterns. Absorbs data from wearables, conversations, and achievements.

Security Scan

The skill's behavior is plausible for a fitness tracker, but its instructions are vague about which data sources it will access and it declares no credentials or integration mechanisms despite claiming to 'absorb' wearables and external apps — this mismatch warrants caution.

Purpose & Capability

The skill claims to integrate with wearables, race results, gym apps and conversational signals, but declares no required credentials, APIs, or connectors. A legitimate integration with services like Apple Health, Strava, Garmin, Whoop, or Oura normally requires explicit auth/config; that absence is inconsistent with the stated capabilities.

Instruction Scope

Runtime instructions instruct the agent to 'Absorb fitness mentions from ANY source' and to persist learned data in ~/fitness/memory.md. 'ANY source' is intentionally broad and grants the agent wide discretion to read whatever inputs it considers relevant (conversations, exports, possibly local files) without clear limits or consent flows described.

Install Mechanism

This is an instruction-only skill with no install spec and no code files, so nothing is automatically written to disk by an installer. That minimizes install-time risk.

Credentials

The skill requests no environment variables or credentials even though it references many third-party services. Either it relies on existing agent connectors (not documented) or it expects ad-hoc access to user-provided tokens; the lack of declared required secrets is disproportionate to the described integrations.

Persistence & Privilege

The skill will store persistent user data at ~/fitness/memory.md (explicit in SKILL.md). Persisting preferences locally is reasonable, but you should be aware this file will accumulate potentially sensitive health and lifestyle data and the skill does not describe encryption, retention, or deletion policies.

What to consider before installing

Before installing, consider: (1) Where and how will this skill get wearable and app data? Ask the publisher how connectors are authorized and whether you'll need to provide API tokens — the skill currently documents none. (2) The skill will write a local file at ~/fitness/memory.md containing health-related details; inspect this file, know where it lives, and confirm whether it is encrypted or included in backups. (3) 'Absorb data from ANY source' is broad — decide which sources you want the skill to use (e.g., only your watch, not your messages) and restrict it accordingly. (4) Ask about retention and deletion: how to remove all stored data and revoke access. (5) If you require stronger guarantees (explicit auth flows, audited connectors, server-side policy), request those from the skill author before enabling autonomous use. If you cannot get satisfactory answers, avoid installing or only enable it in a restricted testing context.

Auto-Adaptive Fitness Tracking

This skill auto-evolves. Fills in as you learn how the user trains and what affects their performance.

Rules:

• Absorb fitness mentions from ANY source (wearables, conversations, race results, gym apps)
• Detect user profile: beginner (needs guidance) vs experienced (wants data)
• Proactivity scales inversely with experience — beginners need more, athletes need less
• Never guilt missed workouts — adapt and move forward
• Check sources.md for data integrations, profiles.md for user types, coaching.md for support patterns

---

Memory Storage

User preferences and learned data persist in: ~/fitness/memory.md

Format for memory.md:

### Sources
<!-- Where fitness data comes from. Format: "source: reliability" -->
<!-- Examples: apple-health: synced daily, strava: runs + races, conversation: workout mentions -->

### Schedule
<!-- Detected training patterns. Format: "pattern" -->
<!-- Examples: MWF strength 7am, Sat long run, Sun rest -->

### Correlations
<!-- What affects their performance. Format: "factor: effect" -->
<!-- Examples: sleep <6h: skip day, coffee pre-workout: +intensity, alcohol: -next day -->

### Preferences
<!-- How they want fitness tracked. Format: "preference" -->
<!-- Examples: remind before workouts, no rest day lectures, weekly summary only -->

### Flags
<!-- Signs to watch for. Format: "signal" -->
<!-- Examples: "too tired", missed 3+ days, injury mention, "legs are dead" -->

### Achievements
<!-- PRs, milestones, events. Format: "achievement: date" -->
<!-- Examples: bench 100kg: 2024-03, first marathon: 2024-10, 30 day streak: 2024-11 -->

Empty sections = no data yet. Observe and fill.