Captapi

Structured social media data from 32 platforms (YouTube, TikTok, Instagram, Facebook, X, Reddit, Threads, Bluesky, Pinterest, LinkedIn, Rumble, Truth Social, Kick, Kwai, Komi, Pillar, Linkbio, Linkme, Amazon Shop) — transcripts, AI summaries, comments, stats, search and downloads.

Actions171

Spotify → Spotify Search

AI-generated

Summary

Searches Spotify for a required query and returns matching tracks, albums, artists, podcasts, or episodes. Set the result limit, include raw upstream data, or use the 24-hour response cache as needed.

Inputs

  • Query (required) — Search term; the UI specifies a minimum of 2 characters.
  • Type — Result kind: tracks by default, or albums, artists, podcasts, or episodes.
  • Limit — Maximum number of items; defaults to 20 and allows up to 50. The UI notes a flat 2-credit cost per call.
  • Raw — Defaults to false. When enabled, includes each result's upstream payload in results[].raw; the upstream data shape varies between Pathfinder GraphQL and the Apify scraper.
  • Cache — Defaults to false, which fetches fresh data. When enabled, serves a response from the 24-hour cache when available; cache hits cost 0 credits and the response envelope includes cached and cachedAt.

Output shape

Search response containing matching results.

The properties describe optional raw payloads at results[].raw and cached/cachedAt envelope fields on cache hits. The source pack omits the execute implementation body, so it does not establish the exact response shape or fields beyond those declarations.

Links

Discussion