Media¶
Everything here is generated from the recorded demo by scripts/launch/make_launch.py, so
it stays in sync with the terminal output. Use it freely when you write or talk about grip.
Launch video¶
84 seconds, 1080p, H.264. Title cards around the full demo. It has a silent audio track, so add music before you upload it anywhere that expects sound.
Demo¶
The same 53 second recording as the GIF on the front page: one commit blocked at 12/100, then accepted at 92/100. Also available as GIF, WebM and an asciinema cast.
Images¶
1270 × 760 gallery image, the size Product Hunt and most link previews want.
240 × 240 icon, also at 480 × 480.
Feature clips¶
One short clip per feature added since launch, and a 94 second cut of all five with title
cards. Each runs the real grip in a real terminal with the offline scripted provider, so
the questions and feedback are canned for the clips' rate-limiter diff; a real model
writes them from your diff.
| Clip | Shows | Length |
|---|---|---|
| grip-whats-new.mp4 | all five, with title cards | 94 s |
| init.mp4 | grip init: config, hook and agent rule files |
11 s |
| agents.mp4 | grip agents: detection and the files it writes |
13 s |
| agent-gate.mp4 | Claude Code's git push denied, then the agent quiz flow |
28 s |
| skip.mp4 | grip skip, a pause, grip statusline, grip resume |
20 s |
| last.mp4 | grip last: your answers, ready for a commit message |
7 s |
Every clip also exists as a GIF next to the MP4.
Regenerating¶
pip install -e ".[demo]"
python scripts/demo/make_demo.py all # re-record and render the demo
python scripts/launch/make_launch.py # rebuild demo.mp4 and everything under assets/launch
python scripts/demo/make_features.py # re-record the feature clips and the stitched cut
The demo script needs bash and a pseudo-terminal. The launch script needs an ffmpeg
with libx264, which the demo extra provides through imageio-ffmpeg.
All three take colours from branding/tokens.json and fonts from Google Fonts (Archivo,
Instrument Sans, JetBrains Mono, cached in build/fonts) through scripts/demo/brandkit.py.
Offline, they fall back to DejaVu and say so.