agentleFS
manthanpatelllmanthanpatelll/leadgenman-video-skills/skills/produce/SKILL.md

produce

Skill17 starsChanged 6 months agoMITView on GitHub

Autonomous rough cut engine. Takes a raw face cam video with retakes and outputs a clean rough cut with all retakes, false starts, and side conversations removed. Uses waveform-first architecture + OpenAI Whisper API. Invoke as /produce with a video file path.

  • Reads credentials

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. openai/plugins★ 7.8ksame name
    2. ttalkkaklab/social-flow★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository5 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Verify file exists: check with ls

    line 35

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Autonomous Rough Cut Engine

Usage

/produce:produce /path/to/raw-video.mp4

What it does

Takes a raw face cam recording (any length, with retakes) and outputs a clean rough cut in original resolution. Removes:

  • Retakes (keeps only the last take of each section)
  • False starts and fragments
  • Side conversations (talking to the AI about the script, not recording)
  • Internal repetitions (e.g., double CTA starts)

Adds:

  • 30ms audio fades at cut points (no clicks/pops)
  • Room tone bed at 12% volume (fills micro-gaps naturally)
  • Smart gap handling (preserves natural breaths between sentences)

Requirements

  • Claude Code
  • Python 3.10+
  • uv (Python package manager)
  • ffmpeg
  • An OPENAI_API_KEY environment variable (used for per-chunk Whisper transcription)

How to run

  1. Get the video path from the user argument
  2. Verify file exists: check with ls
  3. Run the pipeline from the bundled engine:
cd <path-to-this-plugin>/engine
OPENAI_API_KEY=sk-... uv run python roughcut.py "<video_path>"
  1. The script will:
    • Extract audio and analyze waveform (0.3s threshold)
    • Transcribe each speech chunk individually via OpenAI Whisper API
    • Find the last complete run-through
    • Detect and remove retakes (rapidfuzz + key word overlap + partial containment)
    • Build the rough cut with audio fades and room tone
    • Cross-verify by re-transcribing the output
    • Open the final file automatically
  2. Report results to the user: duration, chunks kept/skipped, any issues detected

Output

  • File saved to: exports/produce/<video_name>_<timestamp>/<video_name>_roughcut.mp4
  • Original resolution (1920x1080 or whatever the source is)
  • CRF 16, AAC 256kbps (near-lossless quality)
  • Ready to import into Descript, Premiere, or any editor for final assembly

Architecture

  • Waveform-first: 10ms RMS energy windows find exact speech boundaries
  • Per-chunk transcription: Each speech chunk sent to OpenAI Whisper API separately (prevents merging retakes)
  • Last-take rule: You record multiple takes, keep repeating until satisfied, then move on. Last take before a topic change = keeper.
  • Three-layer retake detection:
    1. Text similarity (rapidfuzz token_sort_ratio > 60)
    2. Key word overlap coefficient > 0.60
    3. Partial containment (short chunk's first words appear in a later chunk)
  • Smart gap handling: Natural breaths between kept chunks preserved, gaps where content was removed get room tone fill

Settings

These defaults are tuned for a natural conversational speaking style. Adjust in engine/roughcut.py if your cadence differs:

  • Silence merge threshold: 0.3s
  • Min chunk duration: 1.0s
  • Fragment max duration for partial containment: 3.0s
  • Breath gap max: 1.5s
  • Room tone volume: 12%

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.