Skip to main content
Glama

CLIPCLIPER MCP server

Give your agent a public video link and get back a timestamped transcript, chapters and clip suggestions. Works with YouTube (videos, shorts, live replays), Twitch VODs, Kick VODs and TikTok, including the platforms that block datacenter IPs.

The server is hosted: nothing to download or run on GPUs. The video is fetched, transcribed with Whisper large-v3 and deleted; only the transcript stays, for 24 hours.

Tools

Tool

What it does

get_transcript

Full transcript with timestamps. Formats: timestamped (default), text, srt, segments (JSON).

get_chapters

Topic chapters: start, end, title, one-line summary.

suggest_clips

The most shareable moments for short-form clips: start, end, hook-style title, reason. Accepts an editorial instruction.

All three take a url. Processing is asynchronous: a call waits up to wait_seconds (default 90) and, if the video is still downloading or transcribing, returns status: "processing". Call the same tool again with the same url; the job keeps running on the server and nothing is charged twice. Results are cached for 24 hours per url.

Related MCP server: Social → Context

Connect

Listed on Smithery (one-click setup for Claude, Cursor and other clients) and submitted to the Glama registry.

Remote (recommended), Streamable HTTP:

https://clipcliper.com/mcp
Authorization: Bearer <your license key>

Claude Code:

claude mcp add --transport http clipcliper https://clipcliper.com/mcp --header "Authorization: Bearer YOUR_KEY"

Claude Desktop, Cursor and other clients that take a JSON config (remote):

{
  "mcpServers": {
    "clipcliper": {
      "type": "http",
      "url": "https://clipcliper.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_KEY" }
    }
  }
}

Local bridge (stdio) for clients that cannot call a URL directly:

{
  "mcpServers": {
    "clipcliper": {
      "command": "npx",
      "args": ["-y", "clipcliper-mcp"],
      "env": { "CLIPCLIPER_LICENSE_KEY": "YOUR_KEY" }
    }
  }
}

The bridge is this repository: a ~100-line process that speaks MCP over stdio and forwards every call to the hosted endpoint with your key. Until the npm package is published you can run it straight from GitHub: npx -y github:Inmoprice/clipcliper-mcp.

Keys and pricing

  • Without a key: each IP can process 3 videos per day of up to 15 minutes. Good for trying it.

  • With a key: minute packs at clipcliper.com/mcp, one-time, minutes never expire:

Pack

Price

Minutes

Per minute

Starter

USD 9

600

USD 0.015

Pro

USD 29

3,000

USD 0.0097

Scale

USD 79

12,000

USD 0.0066

Every video costs one pack minute per started minute of video; chapters and clip suggestions on a transcribed video are free. The license key is emailed to you right after checkout. Send it as Authorization: Bearer <key> (or X-License-Key); it never goes in tool arguments or URLs. Hour packs of the CLIPCLIPER desktop app also work here (one pack hour per started hour).

REST

Prefer plain HTTP? The same pipeline is available at POST https://clipcliper.com/api/link/transcribe, /api/link/chapters and /api/link/suggest with a JSON body {"url": "..."} and the same header. 202 while processing, 200 when done.

Privacy

  • The video file is deleted as soon as the transcript exists; the transcript expires after 24 h.

  • No accounts, no cookies. Usage is counted per key (hours) or per IP hash (free tier).

  • Questions: support@clipcliper.com

License

MIT for the code in this repository (the bridge). The hosted service is subject to clipcliper.com's terms.

Available Tools

3 tools
get_chaptersGet chaptersA
Read-only
Inspect

Split a public video into topic chapters (start/end seconds, title, one-line summary) from its transcript. Transcribes the video first if needed (same cost and async behaviour as get_transcript); the chapters themselves are free and cached.

Inputs: url (required), language, max_chapters, wait_seconds, +1 more.

get_transcriptGet transcriptA
Read-only
Inspect

Download a public video (YouTube, Twitch VOD, Kick VOD, TikTok) and return its full transcript with timestamps (Whisper large-v3). Asynchronous: if it returns status 'processing', call again with the same url. Results are cached 24 h per url. With a license key it costs one pack minute per started minute of video (minute packs at https://clipcliper.com/mcp; desktop hour packs cost one hour per started hour); without a key, videos up to 15 min (daily limit per IP).

Inputs: url (required), format, language, max_chars, +1 more.

suggest_clipsSuggest clipsA
Read-only
Inspect

Find the most shareable moments of a public video for short-form clips: start/end seconds aligned to speech, a hook-style title in the video's language and why it works. Transcribes the video first if needed (same cost and async behaviour as get_transcript). Suggestions are free per call.

Inputs: url (required), count, language, instruction, +4 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.3/5.0

    Scored across 3 tools

    Disambiguation5/5

    Each tool has a distinct output: full transcript vs. topic chapters vs. shareable clip suggestions. Overlap in video preprocessing is clearly explained in descriptions, so an agent can reliably select the right tool.

    Naming Consistency5/5

    All tool names follow a consistent verb_noun pattern: get_transcript, get_chapters, suggest_clips. The prefix 'get_' versus 'suggest_' matches the action precisely, with no mixed naming styles.

    Tool Count5/5

    Three tools is a well-scoped set for a video transcription and clip suggestion server. Each tool earns its place and covers a distinct stage of the workflow without redundancy.

    Completeness4/5

    The surface covers the core workflow of transcribing a video, generating chapters, and finding clip moments. Minor gaps exist (e.g., no video metadata retrieval or actual clip export), but the primary use case is fully supported.

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Transforms social media links (TikTok, Instagram, YouTube) and videos into text context with transcription, OCR, and metadata for AI agents like Claude and Cursor.
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to turn video links from YouTube, TikTok, Instagram, X and Vimeo into usable context: transcripts, summaries, chapters, and searchable answers with timestamped links to the exact moment. All tools are read-only, and the hosted server needs no account or API key to start.
      1
      MIT