agentleFS
bookchaowalitbookchaowalit/platform-agent-skills/flow-video/SKILL.md

flow-video

Skill0 starsChanged 8 days agoApache-2.0View on GitHub

Record read-only walkthrough videos of a running web app — per-role tours that map every page a role sees, and scripted use-case videos with Thai captions. Use when asked for a demo, walkthrough, screen recording, flow or use-case video, a "what does each role see" map, or onboarding and handover material. Not for UX auditing (ux-review), test strategy (qa-testing) or designing flows (product-management).

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    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
  1. Very specific:

    Run source ~/ai-stack/ui-test-env.sh first on this WSL machine.

    line 24
  2. Very specific:

    Add --only UC-02 to redo one video.

    line 30
  3. Distinctive:

    Use demo data, and keep internal IDs, admin pages and other clients out of frame.

    line 63

≈ “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

Flow Video

Show the product as it really behaves, one role and one use case at a time. A video is evidence of what exists today, not a design: record only routes and actions that are live in the environment you point at.

Workflow

  1. Map roles and use cases. Take them from the system's docpack (ux-flows, srs use cases), ux-review P1–P2 output, or the code (role enums, route guards, menu config). Give each use case an ID that matches the docs (UC-01, FR-03). List per role: who they are, what they come to do, and the 2–5 use cases that matter most. Ask the owner only for business names the code does not show.
  2. Pick the environment. Local or staging for anything that writes. Production only with review accounts, read-only, under ../ux-review/references/production-test-accounts.md.
  3. Sign in each role once with ../qa-testing/scripts/ui_login.mjs. Sessions live in ~/.local/state/ui-review/sessions/<app>-<role>.json, never in a repository. Run source ~/ai-stack/ui-test-env.sh first on this WSL machine.
  4. Write the scenario file (references/scenario-format.md): one tour per role first, then one steps scenario per use case. Each step's say is one short Thai sentence about what the user does and why, not a description of the UI. Look up real selectors on the rendered page (role, visible text, href). Do not guess them.
  5. Record: node scripts/flow_video.mjs --scenarios flows.json --out <dir>. Add --only UC-02 to redo one video. The gallery keeps the others. Use --slow 1.3 for a client audience and --mp4 when it will be sent on LINE or by email.
  6. Review every video yourself. Open each <id>/NN.png storyboard frame. Check that the caption matches what is on screen, that the target is highlighted, that no step landed on an error, empty state or sign-in page, and that no personal or client data shows that the audience must not see. A failed scenario writes failed.png and exits 1. Fix the selector or the flow and record it again.
  7. Check the page map (page-map.md) against the intended permissions. A page in a role's menu that refuses (✗) and a page missing from a role that should have it are findings. Log them through ux-review or qa-testing.
  8. Publish inside the boundary. Link the gallery and page map from the system's docpack ux-flows doc (use case ID → video file). Keep the videos out of git (large binaries, they may show data). Store them in the evidence folder, or in the client folder for client work.
  9. Re-record when a flow changes. A release that changes a recorded flow or a role's menu re-records the affected videos in the same change, so the videos never describe an older app.

Decision rules

  • Writes: a scenario that must create or change data sets "writes": true, and the file sets "readOnly": false. The command then needs --allow-writes, and the run must target local or staging. On production that requires the owner's written approval of that exact flow, plus a clean-up.
  • Blocked writes: in read-only mode only GET, HEAD and OPTIONS reach the server, and WebSockets are mocked. blockedWrites lists what was stopped. A step that submits a form ends the scenario as blocked. A use case whose point is a write (submit, approve) is recorded up to the button, with an expect step ("this button sends it for approval"). Do not fake the result.
  • Expired session: a scenario that lands on the sign-in page stops with an error rather than record the public site under the role's name. Sign in again.
  • Audience edition: for a client, show only that client's roles and use cases. Use demo data, and keep internal IDs, admin pages and other clients out of frame. Sending a video outside is an external action and needs approval.
  • No narration audio: captions are burned in and also exported as .vtt. Voice-over is a separate, owner-approved step.
  • Videos are not a review: they show the flow. They do not sign it off. UX findings still go through ux-review and qa-testing.

Resources

  • scripts/flow_video.mjs records tours and use cases. It writes .webm/.mp4, .vtt, step frames, manifest.json, page-map.md and index.html.
  • references/scenario-format.md covers the JSON format, step actions, selectors, pacing and examples. Read it before writing a scenario file.
  • ../qa-testing/scripts/ui_login.mjs saves a role session.
  • ../ux-review/references/production-test-accounts.md sets the rules for review accounts on production.

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.