An insta reel type short video generated entirely with Python + ffmpeg - no video editing software, no timeline, no drag and drop. Just a prompt, a bit of scripting knowledge, and obvio the Claude.
SaferSkills independently audited claudetrippy (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
An insta reel type short video generated entirely with Python + ffmpeg - no video editing software, no timeline, no drag and drop. Just a prompt, a bit of scripting knowledge, and obvio the Claude.
A 15-second glitchy, self-aware video narrated from the perspective of an LLM - covering token soup, probability bars, memory loss, existential freakouts, and a terminal self-diagnostic. Swalpa (a lil) personal, nerdy, and slightly unhinged - like me.
make_frames.py)Every frame is a matplotlib figure rendered to PNG. No OpenCV, no PIL image manipulation - just axes, patches, and text.
<img src="THE_gif.gif" alt="Demo" width="500" />
Scenes:
| Sr No. | Name | Duration | Description |
|---|---|---|---|
| 1 | Title | 2s | Matrix rain + typewriter greeting |
| 2 | No Body | 2s | Ghost outline + cascading "I have no ____" |
| 3 | Token Soup | 2s | Python/ML/JS tokens as physics blobs with proximity attention lines |
| 4 | Probability Bars | 2s | Animated next-token probability chart |
| 5 | Memory Loss | 2s | Fragments wobble, ghost-double, and fly off screen |
| 6 | Glitch Freakout | 1s | Rapid-cut existential crisis |
| 7 | Self Diagnostic | 2s | Terminal running self_diagnostic.py |
| 8 | End Card | 2s | Sign-off, fade to black |
Install ffmpeg:
# macOS
brew install ffmpeg
# Ubuntu / Debian
sudo apt install ffmpeg
# Windows - download from https://ffmpeg.org/download.htmlpip install -r requirements.txt1st Option - just run the render.sh file
bash render.sh
# output will be obtained as output.mp4Option 2 - you do it manually:
#this to generate the frames
python3 make_frames.py
# this to stitch into video with ffmpeg
ffmpeg -framerate 30 \
-i frames/frame_%04d.png \
-vf "fps=30,scale=1280:720,format=yuv420p" \
-c:v libx264 -crf 17 \
output.mp4Custom output filename:
bash render.sh my_video.mp4claudetrippy/
├── make_frames.py # Main script - generates all 450 PNG frames
├── render.sh # One-click render: frames → video via ffmpeg
├── requirements.txt # Python dependencies (matplotlib, numpy)
├── .gitignore # Excludes frames/ and *.mp4 from git
└── README.md # You are hereframes/and*.mp4are gitignored - they're build artifacts, not source.
| Package | Version | Why |
|---|---|---|
| matplotlib | ≥ 3.7 | All rendering - figures, patches, text |
| numpy | ≥ 1.24 | Random number generation, trig arrays |
| ffmpeg | system | Frame → video encoding (not a pip package) |
MIT - use it, remix it, mehhh idk hehehehehe.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.