Quickstart

1. Connect an app

Open Connections and press Connect on the app you want. You approve on the vendor’s own screen, and Selat keeps the tokens and refreshes them. Nothing is stored in your agent.

2. Mint a credential

Open Credentials and create one. It starts with slt_live_ and it is shown once, so copy it then. Selat stores only its hash.

This credential is the only secret your agent holds. Connecting or disconnecting an app later does not change it.

3. Call a tool

Over MCP, which is what most agent runtimes speak:

curl -s -X POST https://api.selat.weekndlabs.com/mcp \
  -H "Authorization: Bearer $SELAT_TOKEN" \
  -H 'content-type: application/json' \
  -H 'accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

Or over REST, if your agent does not speak MCP:

curl -s -X POST https://api.selat.weekndlabs.com/v1/tools/github__get_authenticated_user/call \
  -H "Authorization: Bearer $SELAT_TOKEN" \
  -H 'content-type: application/json' -d '{}'

{"content":{"login":"fajarhide","name":"Fajar Hidayat","public_repos":16},
 "nextCursor":null,"hasMore":false,"request_id":"a8ff2ce9-..."}

Where to go next

MCP clients has the exact configuration for Claude Code, Cursor and any client that speaks Streamable HTTP. Errors and limits covers the codes worth branching on and the tool budget.