Recipes
Copyable workflows for brand monitoring, competitor tracking, transcripts, ads, creator scoring, and Reddit research — with credit callouts and TypeScript SDK examples
Recipes are short, opinionated workflows built from routes that already exist in the API reference. Each one shows which endpoints to call, how credits usually add up, and a TypeScript SDK sketch you can paste into a worker.
They are not a second API. If a route is not linked here, it is not part of the recipe — check /llms-endpoints.txt or the platform pages before inventing one.
Credits in every recipe
Metered lookups bill on completed attempts, including not_found / private on HTTP 200. Routing via POST /v1/ask is free; the nested lookup is not. Always trust meta.creditsCharged. See Credits.
Pick a recipe
Brand monitoring
Ask once to discover routes, then pin platform search on a schedule.
Competitor profile tracking
Fan out the same brand across TikTok, Instagram, YouTube, and X.
Video transcription
YouTube, TikTok, and Instagram transcripts into one worker shape.
Facebook Ad Library
Keyword search plus company ad lists for competitive creative.
Creator engagement scoring
Profile + recent posts, then score median views yourself.
Reddit research
Global search, subreddit scope, and comment pulls for a listening pulse.
How to use these
- Start from Quickstart or the TypeScript SDK if you do not have a key yet.
- Prefer typed routes once you know the platform. Use
POST /v1/ask(or MCPnl_ask_post) only to discoverdata.routedOperation, then pin that method. - In Cursor or Claude, connect MCP and optionally install Agent Skills so the agent reads
/llms.txtinstead of guessing paths. - For longer tutorials with storage and alert patterns, see the Guides section — recipes stay short on purpose.
Related
- Platform hubs — product pages per network
- API reference — generated schemas and examples
- Errors —
lookupStatusvs HTTP status - Credits — metering, balance, and
402