Examples

Structured output for video parsing agents.

Use these examples when writing agent prompts, MCP directory listings, internal docs or tests around public share-link parsing.

Install first LLM context

clijson

$ fast-download parse "https://www.tiktok.com/@example/video/123" --json

Normalized response

The API and MCP tools keep output predictable so agents can reason over media entries and quota state.

responseapplication/json

{
  "platform": "tiktok",
  "title": "@example video",
  "author": "example",
  "media": [
    {
      "type": "video",
      "url": "https://...",
      "watermark": false,
      "quality": "hd",
      "ext": "mp4"
    }
  ],
  "quota": {
    "plan_id": "pro",
    "daily_limit": 1000,
    "used": 1,
    "remaining": 999
  }
}

Agent routing examples

Fastdown's MCP tool descriptions intentionally separate local downloads from metadata-only parsing.

Download

"Save this video locally"

The agent should call download_video with the public share link.

Inspect

"Give me the media URL"

The agent should call parse_video and return the structured media entries.

Limits

"How many parses are left?"

The agent should call quota or run fast-download quota --json.