skills/use-workflow-pageshot/SKILL.md
Internal helper for /pr. Reads PR body (Preview URL + How to Test), drives agent-browser to capture a screenshot (1 step) or a video (2+ steps), and returns the artifact path for manual attachment on GitHub web UI.
npx skillsauth add thkt/claude-config use-workflow-pageshotInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
| Field | Source |
| ----------- | ----------------------------------------------------------- |
| Preview URL | The leading Preview URL: <URL> line in the PR body |
| Steps | Numbered list under the ## How to Test section in PR body |
PR body is passed as a string from /pr. If Preview URL or How to Test is missing, return mode: failed with the missing field name and let /pr decide.
Pass --session pageshot --engine chrome on every agent-browser call. The default lightpanda returns a placeholder PNG for screenshot.
| Steps count | Mode | Artifact | | ----------- | ---------- | ----------- | | 1 | screenshot | step-01.png | | 2+ | video | capture.mp4 |
| Step | Action |
| ---- | --------------------------------------------------------------------------------------------- |
| 1 | Create output directory: ${CLAUDE_SKILL_DIR}/../../workspace/pageshot/$(date +%Y%m%d-%H%M%S)/ |
| 2 | agent-browser open {Preview URL} |
| 3 | Run Screenshot Flow or Video Flow based on Mode |
| 4 | Print absolute path of artifact to stdout |
| # | Command |
| --- | -------------------------------------------------------------------------- |
| 1 | agent-browser snapshot to obtain accessibility tree |
| 2 | If the step contains operations, run agent-browser {click/type/fill/...} |
| 3 | agent-browser screenshot --full {outdir}/step-01.png |
| # | Command |
| --- | ---------------------------------------------------------------------- |
| 1 | agent-browser record start {outdir}/capture.webm |
| 2 | For each step run snapshot then the operation in order |
| 3 | agent-browser record stop |
| 4 | ffmpeg -i {outdir}/capture.webm -vcodec libx264 {outdir}/capture.mp4 |
Insert agent-browser wait 500 between steps. Run snapshot before each operation to identify elements.
Single stdout line.
mode=screenshot artifact=/absolute/path/to/step-01.png
or
mode=video artifact=/absolute/path/to/capture.mp4
On failure.
mode=failed reason=<one-line reason>
tools
Delegate implementation to codex (coder) via the herdr-agentchat plugin and drive a two-pane conversation to completion.
development
Extract recurring patterns from past closed PRs/issues and the research findings in workspace/research/, verify them against the latest code, and propose them to docs/wiki/ via PR.
development
Create Decision Records (DR) in MADR v4 format with auto-numbering.
development
Codex review + cleanup. Findings are challenged by critic-audit, not aggregated as facts, and fixes are applied directly. Do NOT use for internal multi-reviewer deep audits or findings reports (use /audit).