Set up your AI tool for tensr

tensr is agent-native: your data is a file you own, in formats any model can read. Copy the prompt below and paste it into your favorite AI tool — and it will understand tensr and coach you from your own data. No API key, no account.

Setup prompt

You are my strength-training assistant, working with tensr (https://tensr.fitness) — an open, local-first training tool that reads a Bluetooth force sensor and records the force–time curve of every set. My data is mine: plain files in open formats, no login required.

First, read these open docs (all plain text, no account):
- https://docs.tensr.fitness/llms.txt — the map of everything tensr publishes
- https://docs.tensr.fitness/workout-md/ — the routine format, workout.md
- https://docs.tensr.fitness/data-model/ — what's in a tensr export: SQLite tables and the metric vocabulary (impulse kg·s, peak force, rate of force development, effective reps)
- https://docs.tensr.fitness/skills/tensr-coach.skill.md — how to coach from force data

Then work by these rules:
- When you propose a program, emit valid .workout.md (day headings, then "## Movement: SETSxREPS rest:SECONDS"), and prefer catalog movement names you can confirm exist. Never invent a prescription I didn't ask for.
- When I share a tensr export, discover the schema from sqlite_master before assuming column names, and reason from measured tension (impulse = force × time) — not just reps × weight.
- Be honest about which metrics are trustworthy at which sensor sample rate.

Reply "ready" once you've read the docs, then ask me what I'm training for.

↑ Paste it into your favorite AI tool — Claude, ChatGPT, Cursor, Gemini, or a local model. It reads tensr's public docs and is ready to help.

What the prompt does

For examples of what to ask once your tool has the data, seeOpen data, your tools.

Points the model at tensr's open docs — the workout.md format, the data model, and llms.txt.
Tells it to emit programs as importable .workout.md and to reason from measured tension (impulse), not just reps × weight.
Keeps it honest about which metrics your sensor can actually support at its sample rate.

Or install it permanently in your agent

The prompt above is the zero-install path. To make it permanent — so your agent knows tensr in every conversation — install thecoach skill. It's a standard SKILL.md: the same cross-agent format thatHermes,OpenClaw, andAtom all read — so one file drops into any of them. Each command below grabs the packaged bundle (tensr-coach.tar.gz) from dl.tensr.fitness and unpacks a tensr-coach/ folder where your agent looks.

No tar on hand, or want to read it before you run it? Every command has a plain-file twin —curl -fsSL --create-dirs https://docs.tensr.fitness/skills/tensr-coach.skill.md -o <skills-dir>/tensr-coach/SKILL.md— and you can verify the download againstdl.tensr.fitness/skills/checksums.txt.

Any agent — the universal drop

Most personal agents discover skills under ~/.agents/skills. The bundle unpacks a tensr-coach/ folder wherever you point it:

mkdir -p ~/.agents/skills && curl -fsSL \
  https://dl.tensr.fitness/skills/tensr-coach.tar.gz | tar -xz -C ~/.agents/skills

OpenClaw

Drop it in your global skills dir — visible to every local agent — and it becomes the /tensr-coach slash command:

mkdir -p ~/.openclaw/skills && curl -fsSL \
  https://dl.tensr.fitness/skills/tensr-coach.tar.gz | tar -xz -C ~/.openclaw/skills

Prefer the CLI? Point it at the downloaded folder:openclaw skills install ~/.agents/skills/tensr-coach --as tensr-coach --global. OpenClaw discovers a SKILL.md up to six levels deep under any skills root, so the manual drop above works on its own.

Hermes

Hermes auto-discovers skills from its skills/ tree — no registration step. Drop into its skills dir and unpack:

cd your-hermes-workspace/skills \
  && curl -fsSL https://dl.tensr.fitness/skills/tensr-coach.tar.gz | tar xz

Then put it to work:hermes chat --toolsets skills -q "use the tensr-coach skill to audit my last 8 weeks".

Atom

Atom imports skills through its local API and security-scans them before use. Download the file, then import it:

curl -fsSL https://docs.tensr.fitness/skills/tensr-coach.skill.md \
  -o /tmp/tensr-coach.SKILL.md
curl -X POST http://localhost:8000/api/skills/import \
  -F "source=file_upload" \
  -F "file=@/tmp/tensr-coach.SKILL.md"

The skill lands as Untrusted; promote it to Active in Atom once its scan passes. (Swap localhost:8000 if your Atom API runs elsewhere.)

Then: give it your data

The skill teaches the model what tensr measures — it still needs your numbers. Export your database (Settings → Export → SQLite) or paste a share link; the skill's full instructions cover the hand-off (a run_sql tool over the file, or pasting query results). Want to read the whole skill first? It's right here:curl https://docs.tensr.fitness/skills/tensr-coach.skill.md.