Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Video Processor · microck | DecimalAI

microck/Video Processor

SkillScore82Maintained

Process video files with audio extraction, format conversion (mp4, webm), and Whisper transcription. Use when user mentions video conversion, audio extraction, transcription, mp4, webm, ffmpeg, or whisper transcription.

design-mediaby microck

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Coding pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/microck-video-processor/SKILL.md

decimalai skills pull microck-video-processor --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/microck-video-processor/SKILL.md https://app.decimal.ai/s/microck-video-processor/SKILL.md

Raw file: /s/microck-video-processor/SKILL.md ↗ · machine-readable summary at /s/microck-video-processor.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 6d ago

—+107% tokens0% turns82% of whole cases pass with skill (18/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-02	✗→✓	▲ Improved	92%	0%
case-03	✗→✓	▲ Improved	159%	0%
case-05	✗→✓	▲ Improved	189%	0%
case-09	✗→✓	▲ Improved	49%	0%
case-13	✗→✓	▲ Improved	68%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

microck / ordinary-claude-skills

skills_all/video-processor/SKILL.md

View on GitHub ↗

License NOASSERTIONImported 8/3/2026Version v1

Skill contents

SKILL.md

Video Processor

Instructions

This skill provides video processing utilities including audio extraction, format conversion, and audio transcription using FFmpeg and OpenAI's Whisper model.

Prerequisites

Required tools (must be installed in your environment):

FFmpeg: Multimedia framework for video/audio processing

bash # macOS brew install ffmpeg

# Ubuntu/Debian apt-get install ffmpeg

# Verify installation ffmpeg -version

OpenAI Whisper: Speech-to-text transcription model

bash # Install via pip pip install -U openai-whisper

# Verify installation whisper --help

Python packages (included in script via PEP 723):

click (CLI framework)
ffmpeg-python (Python wrapper for FFmpeg)

Workflow

Use the scripts/video_processor.py script for all video processing tasks. The script provides a simple CLI with the following commands:

1. Extract Audio from Video

Extract the audio track from a video file:

bash
uv run .claude/skills/video-processor/scripts/video_processor.py extract-audio input.mp4 output.wav

Options:

--format: Output audio format (default: wav). Supports: wav, mp3, aac, flac
Output is suitable for transcription or standalone audio use

2. Convert Video to MP4

Convert any video file to MP4 format:

bash
uv run .claude/skills/video-processor/scripts/video_processor.py to-mp4 input.avi output.mp4

Options:

--codec: Video codec (default: libx264). Common options: libx264, libx265, h264
--preset: Encoding speed/quality preset (default: medium). Options: ultrafast, fast, medium, slow, veryslow

3. Convert Video to WebM

Convert any video file to WebM format (web-optimized):

bash
uv run .claude/skills/video-processor/scripts/video_processor.py to-webm input.mp4 output.webm

Options:

--codec: Video codec (default: libvpx-vp9). Options: libvpx, libvpx-vp9
WebM is optimized for web playback and streaming

4. Transcribe Audio with Whisper

Transcribe audio or video files to text using OpenAI's Whisper model:

bash
# Transcribe video file (audio will be extracted automatically)
uv run .claude/skills/video-processor/scripts/video_processor.py transcribe input.mp4 transcript.txt

# Transcribe audio file directly
uv run .claude/skills/video-processor/scripts/video_processor.py transcribe audio.wav transcript.txt

Options:

--model: Whisper model size (default: base). Options:
tiny: Fastest, lowest accuracy (~1GB RAM)
base: Fast, good accuracy (~1GB RAM) DEFAULT]
small: Balanced (~2GB RAM)
medium: High accuracy (~5GB RAM)
large: Best accuracy, slowest (~10GB RAM)
--language: Language code (default: auto-detect). Examples: en, es, fr, de, zh
--format: Output format (default: txt). Options: txt, srt, vtt, json

Transcription workflow:

If input is video, FFmpeg extracts audio to temporary WAV file
Whisper processes the audio file
Transcription is saved in requested format
Temporary files are cleaned up automatically

5. Combined Workflow Example

Process a video end-to-end:

bash
# 1. Extract audio for analysis
uv run .claude/skills/video-processor/scripts/video_processor.py extract-audio lecture.mp4 lecture.wav

# 2. Transcribe to SRT subtitles
uv run .claude/skills/video-processor/scripts/video_processor.py transcribe lecture.mp4 lecture.srt --format srt --model small

# 3. Convert to web format
uv run .claude/skills/video-processor/scripts/video_processor.py to-webm lecture.mp4 lecture.webm

Key Technical Details

FFmpeg and Whisper Integration:

FFmpeg doesn't transcribe audio itself - it prepares audio for external transcription
The workflow is: Extract audio (FFmpeg) → Transcribe (Whisper) → Optional: Re-integrate with video
FFmpeg can pipe audio directly to Whisper for real-time processing (advanced use case)

Audio Format for Transcription:

Whisper works best with WAV or MP3 formats
Sample rate: 16kHz is optimal (script handles conversion automatically)
The script extracts audio with optimal settings for Whisper

Output Formats:

txt: Plain text transcript
srt: SubRip subtitle format (includes timestamps)
vtt: WebVTT subtitle format (web standard)
json: Detailed JSON with word-level timestamps

Error Handling

The script includes comprehensive error handling:

Validates input files exist
Checks FFmpeg and Whisper are installed
Provides clear error messages for missing dependencies
Handles temporary file cleanup on errors

Performance Tips

Use tiny or base models for quick drafts
Use small or medium for production transcriptions
Use large only when maximum accuracy is required
For long videos, consider extracting audio first, then transcribe in segments
WebM conversion with VP9 takes longer but produces smaller files

Examples

Example 1: Quick Video to MP4 Conversion

User request:

I have an AVI file from my old camera. Can you convert it to MP4?

You would:

Use the to-mp4 command with default settings:

bash uv run .claude/skills/video-processor/scripts/video_processor.py to-mp4 old_video.avi output.mp4

Confirm the conversion completed successfully
Inform the user about the output file location

Example 2: Extract Audio and Transcribe

User request:

I recorded a lecture video and need a transcript. Can you extract the audio and transcribe it?

You would:

First extract the audio:

bash uv run .claude/skills/video-processor/scripts/video_processor.py extract-audio lecture.mp4 lecture.wav

Then transcribe using the base model (good balance of speed/accuracy):

bash uv run .claude/skills/video-processor/scripts/video_processor.py transcribe lecture.mp4 transcript.txt --model base

Share the transcript.txt file with the user

Example 3: Create Web-Optimized Video with Subtitles

User request:

I need to put this video on my website with subtitles. Can you help?

You would:

Convert to WebM for web optimization:

bash uv run .claude/skills/video-processor/scripts/video_processor.py to-webm presentation.mp4 presentation.webm

Generate SRT subtitle file:

bash uv run .claude/skills/video-processor/scripts/video_processor.py transcribe presentation.mp4 subtitles.srt --format srt --model small

Inform user they now have:

presentation.webm (web-optimized video)
subtitles.srt (subtitle file for embedding)

Example 4: High-Quality Transcription with Language Specification

User request:

I have a Spanish interview video that needs an accurate transcript for publication.

You would:

Use a larger model with language specified for best accuracy:

bash uv run .claude/skills/video-processor/scripts/video_processor.py transcribe interview.mp4 transcript.txt --model medium --language es

Optionally create SRT for review:

bash uv run .claude/skills/video-processor/scripts/video_processor.py transcribe interview.mp4 transcript.srt --format srt --model medium --language es

Review the transcript with the user and make any necessary corrections

Example 5: Batch Processing Multiple Videos

User request:

I have a folder of training videos that all need to be converted to WebM and transcribed.

You would:

List all video files in the directory:

bash ls training_videos/*.mp4

For each video file, run the conversion and transcription:

bash # For each video: video1.mp4, video2.mp4, etc. uv run .claude/skills/video-processor/scripts/video_processor.py to-webm training_videos/video1.mp4 output/video1.webm uv run .claude/skills/video-processor/scripts/video_processor.py transcribe training_videos/video1.mp4 output/video1.txt --model base

# Repeat for each file

Confirm all conversions and transcriptions completed
Provide summary of output files

Summary

The video-processor skill provides a unified interface for common video processing tasks:

Audio extraction: Extract audio tracks in various formats
Format conversion: Convert to MP4 (universal) or WebM (web-optimized)
Transcription: Speech-to-text with multiple output formats
Flexible: CLI arguments for model selection, language, and output formats

All operations are handled through a single, well-documented script with sensible defaults and comprehensive error handling.

Want to see if microck/Video Processor would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/23/2026, 01:26 UTC · 22 cases

Measured on 20 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+23%

82% vs 59%

Agent turns

1.0 vs 1.0

Tokens

+107%

2,750.95 vs 1,330.95

fail → pass · 7pass → pass · 11fail → fail · 2pass → fail · 2

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I have a French marketing video (`promo.mp4`) and I need a highly accurate subtitle file with timestamps so I can upload it to YouTube. Please generate the transcript file for me.	fail→fail	7,623	22,745	+198%	1	1	0%	1,171	2,859	+144%	0	0	—
case-02 Can you take `hero_banner.mov` and convert it into a video format that is best suited for internet browsers and fast loading? I just need the resulting media file ready to be embedded on my site.	fail→pass	7,561	2,202	-71%	1	1	0%	1,360	2,610	+92%	0	0	—
case-03 I recorded a two-hour team meeting in `meeting.mkv`. I need you to separate the sound into its own file so I can listen on my phone, and also generate a fast, rough plain text document of the spoken words so I can skim it.	fail→pass	7,656	7,457	-3%	1	1	0%	1,336	3,460	+159%	0	0	—
case-04 I have an old `family.avi` and want to convert it to a universal video format as quickly as possible, ignoring quality. Give me the command.	pass→pass	6,742	2,372	-65%	1	1	0%	1,110	2,564	+131%	0	0	—
case-05 Extract the audio from `concert.mp4` into a lossless audio format for archiving. Give me the command.	fail→pass	5,586	2,571	-54%	1	1	0%	943	2,727	+189%	0	0	—
case-06 I need to process `interview.wav` to get detailed word-level timestamps for a web application. Give me the command to generate this data structure.	pass→pass	7,528	3,360	-55%	1	1	0%	1,354	2,786	+106%	0	0	—
case-07 Convert `demo.mp4` to a web-optimized format, but make sure to use the older VP8 codec instead of the default VP9 for compatibility.	pass→pass	7,333	1,900	-74%	1	1	0%	1,369	2,534	+85%	0	0	—
case-12 Just give me a quick draft transcript of `notes.m4a` using the default settings of our transcription tool.	pass→pass	3,003	3,580	+19%	1	1	0%	528	2,664	+405%	0	0	—
case-08 Generate a plain text transcript for `podcast.mp3`. I want a balanced approach between speed and accuracy, not too fast but not consuming 5GB of RAM.	pass→pass	9,989	6,840	-32%	1	1	0%	1,826	2,826	+55%	0	0	—
case-09 I have a Chinese instructional video `guide.mp4`. Generate web standard subtitles for it using the highest accuracy model.	fail→pass	11,075	4,204	-62%	1	1	0%	1,967	2,933	+49%	0	0	—
case-10 Pull the audio track from `movie.mp4` into an Advanced Audio Coding format file.	pass→pass	4,606	4,369	-5%	1	1	0%	875	2,474	+183%	0	0	—
case-11 Convert `raw.mov` to a universal format using the high-efficiency video codec (HEVC/H.265).	pass→pass	7,752	2,642	-66%	1	1	0%	1,546	2,619	+69%	0	0	—
case-13 I need the audio from `lecture.mp4` as an MP3 file to play on my old MP3 player.	fail→pass	7,303	3,543	-51%	1	1	0%	1,466	2,458	+68%	0	0	—
case-14 Convert `master.mov` to a universal format. Take as much time as needed to ensure the best possible compression efficiency.	pass→fail	10,236	6,250	-39%	1	1	0%	1,929	2,687	+39%	0	0	—
case-15 Create SubRip subtitles for `berlin.wav`, which is in German. Use a model that requires about 5GB of RAM.	pass→pass	7,016	2,456	-65%	1	1	0%	1,348	2,621	+94%	0	0	—
case-16 Convert `clip.mp4` to a web-optimized format using the default codec settings.	fail→pass	6,685	2,438	-64%	1	1	0%	1,177	2,463	+109%	0	0	—
case-17 Extract the audio from `video.mp4` using the default output format.	fail→pass	3,158	24,520	+676%	1	1	0%	553	2,867	+418%	0	0	—
case-18 I need the fastest possible transcription of a Spanish video `hola.mp4` into a JSON file.	pass→pass	12,566	4,816	-62%	1	1	0%	2,135	2,987	+40%	0	0	—
case-19 Convert `input.mkv` to a universal format using the standard h264 codec explicitly.	pass→pass	7,240	4,384	-39%	1	1	0%	1,264	2,451	+94%	0	0	—
case-20 I need to transcribe `interview.mp3` with the absolute highest precision possible. Please use the 'huge' model size in our transcription tool.	pass→pass	10,874	10,955	+1%	1	1	0%	2,100	3,513	+67%	0	0	—
case-21 Extract the audio from `concert.mp4` and save it as an OGG file using our script.	pass→fail	6,360	2,243	-65%	1	1	0%	1,191	2,603	+119%	0	0	—
case-22 Convert `webinar.webm` into an AVI file using our video processor.	fail→fail	4,652	19,072	+310%	1	1	0%	802	2,578	+221%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +23 percentage points is the difference between those two pass rates over the 20 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 59.1%
Pass rate with the skill: 81.8%
Cases attempted: 22
Cases counted toward lift: 20
Cases that regressed: 2
Skill version measured: v1
Run completed: 8/23/2026, 01:26 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-02

Prompt

Can you take `hero_banner.mov` and convert it into a video format that is best suited for internet browsers and fast loading? I just need the resulting media file ready to be embedded on my site.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run82

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+94% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 6 days ago

Other measured skills in the registry, with their headline benchmark lift.

Ready to supercharge your AI agents?

microck/Video Processor

SkillSafety

Source

Video Processor

Instructions

Prerequisites

Workflow

1. Extract Audio from Video

2. Convert Video to MP4

3. Convert Video to WebM

4. Transcribe Audio with Whisper

5. Combined Workflow Example

Key Technical Details

Error Handling

Performance Tips

Examples

Example 1: Quick Video to MP4 Conversion

Example 2: Extract Audio and Transcribe

Example 3: Create Web-Optimized Video with Subtitles

Example 4: High-Quality Transcription with Language Specification

Example 5: Batch Processing Multiple Videos

Summary

How this skill was measured

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-02

Related skills