Matching quickstart

Matching turns the names people use ("db bench", "rdl") into Open Exercise IDs, for an app with nobody signed in: say, to map the exercise list in your own app once. It's Lifterclub's AI service, not part of the core, so it runs against the sandbox or production, not the template server.

Your app uses a token of its own (the client credentials grant), which reaches matching and nothing of anyone's.

1. Register a confidential client

Once, and keep the secret:

curl -s https://lifterclub-api.runmighty.dev/oauth/register \
  -H 'Content-Type: application/json' \
  -d '{"client_name": "My App", "grant_types": ["client_credentials"], "token_endpoint_auth_method": "client_secret_basic"}'

The answer has client_id (lcc_…) and client_secret (lcs_…).

2. Take a token

curl -s https://lifterclub-api.runmighty.dev/oauth/token \
  -u "$CLIENT_ID:$CLIENT_SECRET" \
  -d grant_type=client_credentials

The access_token lasts an hour and carries only catalogue:match. There's no refresh token: ask again.

3. Match

curl -s https://lifterclub-api.runmighty.dev/v1/match \
  -H "Authorization: Bearer $TOKEN" \
  -H 'Content-Type: application/json' \
  -d '{"names": ["db bench", "rdl", "flurbo lift"]}'

Each name comes back in order with a status: catalogue is sure, close is likely (let your user confirm it), unmatched has no exercise. candidates lists the alternatives, best first. Up to 50 names a call; one call counts once against your app's budget. Nothing is stored.

From TypeScript or anything else, it's one request:

const res = await fetch('https://lifterclub-api.runmighty.dev/v1/match', {
  method: 'POST',
  headers: { Authorization: `Bearer ${appToken}`, 'Content-Type': 'application/json' },
  body: JSON.stringify({ names: ['db bench', 'rdl'] }),
})
const { matches } = await res.json()

Assistants have the same matching as the MCP tool match_exercises.