CLI-Native Screen Recording Makes Demos Scriptable

articletoolclimacosvideoscreen-recordingagent-workflowsjoelclaw

CLI-controlled recording maps to agent-produced proof videos for joelclaw demos and workflow receipts.

Screen recording usually lives behind a heavy GUI, which makes it awkward for agents, scripts, and repeatable demos. Steve Tenuto introduced Framecap as a macOS screen recorder you run from the terminal: tracks in, files out, no project timeline ceremony.

The clever bit is that recording becomes scriptable infrastructure. Framecap can record multiple sources into separate files, hide specific apps, capture a window with a transparent background, and write Apple ProRes or H.264. The docs expose it as framecap -t ..., with track specs for screen, camera, audio, filters, app targeting, filenames, and output dirs.

For joelclaw, this is useful where a run should leave evidence behind: CLI walkthroughs, release receipts, UI bug captures, and demo videos that prove the thing actually happened. A shell-native recorder means the proof can be part of the workflow instead of a human afterthought.

Key Ideas

  • Framecap turns macOS screen recording into a CLI command instead of a GUI-only ritual.
  • Multi-track capture can write separate files for screen, camera, and audio, which is cleaner for editing and audit trails.
  • Window-only recording with transparent background makes polished product demos less dependent on manual cropping.
  • App hiding is a practical privacy feature for agent/operator workflows where Slack, terminals, or secrets-adjacent windows might be open.
  • Track specs, filters, filenames, and output directories make recordings reproducible enough to wire into scripts.