A training copilot, gone AI-native. feezify reads your day — your objective form (training load) crossed with how you actually feel — and tells you, in plain words, whether today is green, amber, or red, and why. It does not prescribe workouts. You decide; it helps you read yourself more clearly.
It installs on your own AI as a skill, reads a folder of markdown you own, and runs entirely on your machine — no server, no account, no telemetry.
🇫🇷 Version française : fr/README.md
Start here
Pick the door that fits you — you don’t need to read everything.
| If you… | Read |
|---|---|
| just want to know what this is | What is feezify? |
| don’t want to touch a terminal | Install — no terminal (3 steps) |
| are a developer / OpenClaw user | Install — developer path |
| are set up and want to use it | Daily use |
| want to understand how it reads you | The method |
| care about your data | Privacy & consent |
| are stuck or have a question | FAQ & troubleshooting |
The whole thing in one minute
- Connect Strava (optional) so feezify can see your training load — via Claude’s official Strava connector (no keys), or your own tokens on the developer path.
- Install the skill on your AI (upload a zip in Claude, or point a dev setup at the
SKILL.md). - Keep a short journal — a few honest lines a day (sleep, fatigue, mood, a niggle).
- Ask “how am I today?” — feezify runs a deterministic read and answers green / amber / red and why. It never prescribes.
Everything below just expands those four steps.
The documents
- What is feezify? — the idea, what it does and does not do, and how the old SaaS became an AI-native agent.
- Installation — both ways in: the no-terminal 3-step path for athletes, and the developer path (clone, build, tokens).
- Daily use — your core folder explained, how to journal, how to ask for the read, and how to read the answer.
- The method — how the read is computed: training load (TSB), readiness, and the crossing rule. Open on purpose.
- Privacy & consent — local-first, consent-gated connectors, where tokens live, and how to revoke everything.
- FAQ & troubleshooting — common questions and fixes for both paths.
Canonical sources in the repo
These docs are the reader’s manual. The source of truth for each subject lives with the code and wins in case of conflict:
- The method —
src/adapters/claude-skill/method.md - The skills (what the AI actually does) —
src/adapters/*/SKILL.md - The onboarding checklists —
onboarding.md·fr/onboarding.md - Agent/security rules —
AGENTS.md
Status & license
Early build — beta, feedback welcome. Open an issue with what reads true and what
doesn’t. Licensed Apache-2.0. Built by Nicolas Jouanno — see
author.md.