DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to use your own LLM key for sharper clip selection with Autoclip

Point highlight detection at Anthropic, OpenAI, Gemini or an OpenAI-compatible endpoint while video stays local.

  • Content
  • Intermediate
  • 45 minutes
  • From the README
  • Any agent

Credentials. Your provider API key is stored by AutoClip and billed per request, so use a key with a spending limit.

Who it is forA video editor whose local model picks weak moments and who already pays for an API key.

The problemSmall local models can miss the moments that actually make good clips.

The outcomeBetter ranked highlights, with only transcript text sent to the provider and never the video or audio.

What you need first

  • Python 3.11 or 3.12 (not 3.13)
  • A full ffmpeg build with libass and libx264
  • Node 20+ to build the UI
  • Ollama for fully local runs, or an Anthropic, OpenAI, Gemini or OpenAI-compatible key

Steps

  1. Install AutoClip and run autoclip doctor
  2. Store your provider key with autoclip config set-secret anthropic
  3. Start autoclip serve and choose the provider in the app
  4. Run the same video again and compare the ranked clips

Prompt for your agent

Configure artbyjazi/autoclip with `autoclip config set-secret anthropic`, re-run highlight detection on <video>, and list the top five clips with timestamps. Verify that no audio or video file was uploaded, only transcript text.

Caveat. Only the transcript text leaves your machine with a hosted provider, so sensitive spoken content still reaches that provider.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for artbyjazi/autoclip and "Use your own LLM key for sharper clip selection"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer