demo-video
Skill13 starsChanged 36 days agoMITView on GitHub
Record, edit, and publish the JustSessions homepage demo video with Scripts/DemoVideo, from an isolated copy of the app with sample sessions and a disposable SSH host. Use when re-recording the demo for a new release, changing its takes, cut, narration, or chapters, or remaking the poster or the social card from it.
- Reads credentials
28 other projects use a version of this.
Adopted by
Most recently taken up by
- ohong/agent-skillstheir own version, first seen Nov 16, 2026
- ohong/skillstheir own version, first seen Nov 15, 2026
- Kewton/CommandMatetheir own version, first seen Nov 13, 2026
- SaudSatopay/bullseyetheir own version, first seen Nov 8, 2026
- dxdleady/playwright-demo-videotheir own version, first seen Nov 1, 2026
- yzhao062/agent-configtheir own version, first seen Nov 1, 2026
Used with
Projects with it also use karpathy-coder, ai-act-readiness and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
csiddhant796-blipSkillkarpathy-coder
Use when writing, reviewing, or committing code to enforce Karpathy's 4 coding principles — surface assumptions before coding, keep it simple, make surgical changes, define verifiable goals. Triggers on "review my diff", "check complexity", "am I overcomplicating this", "karpathy check", "before I commit", or any code quality concern where the LLM might be overcoding.
Also in 4 projects that use this
Hussain-ElfalahSkillai-act-readiness
Also in 3 projects that use this
Hussain-ElfalahSkillc-level-agents
Also in 3 projects that use this
csiddhant796-blipSkillself-eval
Honestly evaluate AI work quality using a two-axis scoring system. Use after completing a task, code review, or work session to get an unbiased assessment. Detects score inflation, forces devil's advocate reasoning, and persists scores across sessions.
Also in 4 projects that use this
- +8 more from Hussain-Elfalah
Other versions28 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 28
- liaohch3/claude-tap★ 3.3ksame name
- microsoft/hve-core★ 1.6ksame name
- alinaqi/maggy★ 711same name
- alinaqi/claude-bootstrap★ 708same name
- getsentry/junior★ 421same name
- hmislk/hmis★ 245same name
- Kewton/CommandMate★ 44same name
- Hassaan146/claude-skills★ 18same name
- yzhao062/agent-config★ 16same name
- mrieck/demoday-claude-plugin★ 12same name
18 more not listed
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 repositoryOne other file it gives its agents.
- AGENTS.mdAGENTS.md
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 0 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 26 edited versions we read keep these.
0
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
Update editing/timeline/scenes.py to the new takes.
- Very specific: line 33
Honesty: Every action is real input to the real app.
“Common” is measured: kept by at least 6 in 10 of the 26 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Demo video
Scripts/DemoVideo/demo-video runs every step. Read docs/development/demo-video.md first: requirements, what each setup step changes on the machine, take order, and teardown. This skill adds the workflow and the judgment calls.
Ask before these
record: it moves the pointer and types on the user's screen, and a take aborts if anything else comes to the front. Ask the user to step away and turn on Do Not Disturb, then run it.setup remoteandteardown: they change the shared test VM (adevuser, its hostname) and~/.ssh/config.reset: it opens the demo app's window on the user's screen.
Everything else (setup app|home|editing|sessions, clicks, voiceover, stills, render, web, poster, window-still) runs without taking over the Mac.
Workflow
make buildat the commit to show, thendemo-video setup. Ifsetup sessionsreports the wrong counts, run the CLI that fell short in the demo home withenvironment/run_in_isolated_tmux.shand read the screen it prints. A first-run notice or a folder-trust prompt that swallows the typed prompt has been the cause; mark it as seen infixtures/cli_configs.py.demo-video record all, thendemo-video clicks. After re-recording take A, updateREFERENCEinediting/analysis/find_click_positions.pyfirst.- Update
editing/timeline/scenes.pyto the new takes. Its times are seconds into a take: a line's timestamp in$DEMO_WORK/takes/<take>.driver.logminusRECORDING STARTEDin<take>.recorder.log. Themarklines name the moments the cut is built around. demo-video voiceoveronly for lines that changed, by id. Kokoro is not deterministic, so regenerating every line changes timing everywhere.demo-video stillsat each scene's key moments; look at every still before rendering. Thendemo-video renderand watch the master with sound.- Publish (below), then
demo-video teardownonce the user is done.
Judgment calls
- Pacing. Moments that matter play at 1×: the click that resumes, the double-click that opens an agent, the drag, typed text. Pointer travel, CLI start-up, and waiting play faster, up to 3.2× in the published cut. Idle time before and after a take is cut.
- Narration. One short sentence per beat, said as the thing happens on screen. Captions come from the narration's sentence timing and are placed on the whole video's clock, so a line that runs past its scene spills its audio and caption into the next one. Keep each line shorter than the footage it describes.
- Camera. Zoom in when the point is small text, such as a sidebar row or a CLI's output; the published cut goes up to 1.85×. Between scenes of the same take, start at the camera the previous scene ended on. Check in stills that the pointer and the thing it acts on stay in frame.
- Sample data. Sessions, projects, and replies are fixtures; change them in
fixtures/andenvironment/canned_model_server.py, never by recording real work. Before publishing, check the stills for real names, account paths, hostnames, or tokens. Everything on screen should be under~/Projectsin the demo home or ondevbox. - Honesty. Every action is real input to the real app. Don't fake a state with an overlay or an edit; change the fixtures or the takes instead. Keep
website/media/README.mdtruthful about what is canned.
Publish
demo-video web website/media/justsessions-demo.mp4anddemo-video poster <seconds> website/media/justsessions-demo.jpg. The poster is a frame with no caption that shows the product at its best; it was the split view at 28.0 seconds.- Chapter times: each chapter's
data-start/data-endinwebsite/index.htmland the table inwebsite/media/README.mdcome from scene starts in$DEMO_WORK/build/timeline.json. - Length: the
0:53label and theVideoObjectname,duration, anduploadDateinwebsite/index.html, and the length indocs/development/website.mdandwebsite/media/README.md. - Provenance in
website/media/README.md: recording date, app version and commit, the remote Claude Code version. - Social card, if the poster moment changed or the app looks different:
demo-video window-still <seconds> docs/images/demo-split-view.jpg, then recapturewebsite/social/social-preview.pngfromwebsite/social/preview.htmlasdocs/development/website.mddescribes, and update the row indocs/images/README.md. - Build the site with
make website-checkand check the homepage. Chapter jumps need byte-range responses, whichpython3 -m http.serverdoesn't send; check them on the published site after deploying. make verify, then follow AGENTS.md for the pull request.
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
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.

