agentleFS
LegendAppLegendApp/legend-skills/video-recording/SKILL.md

video-recording

Skill38 starsChanged 41 days agoMITView on GitHub

Record running app flows, animations, and bug reproductions as videos, with continuous capture and verification of the exported motion timing. Use when producing or evaluating app recordings; use screenshot tooling for still images.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Verify capture permission and available disk space before starting.

    line 28
  2. Distinctive:

    Start recording and await the recorder's ready signal before navigation.

    line 35
  3. Distinctive:

    Keep the original recording.

    line 43
  4. Distinctive:

    Read verification.md. Run the verifier on the final exported video using intervals where motion is expected, and source timings when …

    line 48
  5. Distinctive:

    Do not report passed verification based only on a successful export or decode.

    line 61
  6. Distinctive:

    If recording is blocked, deliver the exact blocker and prepared command or flow instead of presenting still captures as a faithful video.

    line 65
  7. Very specific:

    Run regression checks from this skill directory: python3 scripts/test-macos-lifecycle.py (macOS) and python3 scripts/test-verify-video.py.

    line 68

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Video Recording

Deliver a playable recording of the actual running app, with evidence that it captures the requested behavior and preserves its timing. Keep recording independent of navigation and inspection tools.

Capture

  • Identify the requested app, window or device, flow, and output location. Reuse an existing app session. Scope capture to that surface; an app with multiple windows needs an exact window selection, not whichever becomes foreground.
  • Choose a continuous recorder. For native macOS, read macos.md and use the window recorder supplied here or an already-working native recorder. For other targets, read platforms.md. Prefer the environment's designated recording tool; do not install software or change permissions without the applicable authorization.
  • Target 60 fps for animation review where supported; 30 fps can suffice for ordinary flows. A target or exported frame-rate label is not a measurement of captured motion. Preserve source sample timestamps when the recorder exposes them. Never assemble UI-tool screenshots into a video to judge animation smoothness: tool waits, focus changes, and request timestamps distort motion.
  • Verify capture permission and available disk space before starting. Honor the session's reserve, monitor at least every 30 seconds during writing, retain process handles, and bound recording duration. Do not delete artifacts without authorization. Stop at permission or capacity failures; surface the blocker rather than silently substituting a low-rate capture.
  • Record a 3–5 second calibration of known continuous motion. Verify it before the full run. Do not accept black frames or a still screenshot as calibration.
  • Start recording and await the recorder's ready signal before navigation. Allow a short lead-in, traverse the flow with deliberate pauses, then allow a tail. Record uninterrupted across tool calls; inspection may wait for UI stability, but the recorder must not. Always stop/finalize, including after an error, and verify that this thread's recorder has stopped.

Verify and deliver

Keep the original recording. Optional edits may trim known setup or settled pauses, but preserve the timing of motion, failures, and transitions under review. Do not interpolate motion, speed up animations, or replace missed frames to conceal a capture failure. Disclose any edits.

Read verification.md. Run the verifier on the final exported video using intervals where motion is expected, and source timings when available:

python3 scripts/verify-video.py flow.mp4 --motion 1.2:2.4 --motion 5.1:6.3 \
  --timings flow.mp4.frames.jsonl --report flow.quality.json

Watch every requested transition at normal speed in a supported video player. A contact sheet establishes coverage, not smoothness. Automated checks flag stalls; they cannot alone establish their cause or certify visual quality. Distinguish observed app behavior, capture defects, and unresolved uncertainty. Do not report passed verification based only on a successful export or decode.

Embed or link the final video. Briefly report duration, actual cadence evidence and limitations, requested flow coverage, edits, and sizable intermediates. If recording is blocked, deliver the exact blocker and prepared command or flow instead of presenting still captures as a faithful video.

Run regression checks from this skill directory: python3 scripts/test-macos-lifecycle.py (macOS) and python3 scripts/test-verify-video.py.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.