The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to navigate to and interact with arbitrary external websites, which serves as a vector for indirect prompt injection if those sites contain malicious instructions for the agent.
Ingestion points: Browser navigation via page.goto(url) as seen in SKILL.md (e.g., recordDemo, recordFormDemo, recordNavDemo) and reference.md (record template).
Boundary markers: Absent. The skill does not provide instructions to the agent to ignore content from the navigated pages.
Capability inventory: File system access (fs.mkdirSync, fs.renameSync), shell command execution (child_process.execSync), and full browser automation via Playwright.
Sanitization: Absent. The external content is rendered and processed without sanitization.
[COMMAND_EXECUTION]: The skill provides utility code that uses child_process.execSync to invoke ffprobe for calculating video metadata.
Evidence: The getVideoDuration function in reference.md executes a shell command using a template string: execSync(ffprobe -v error -show_entries format=duration -of csv=p=0 "${filePath}").
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the Playwright framework and its browser binaries from the official registry.
Evidence: npm install -D playwright @playwright/test and npx playwright install chromium in SKILL.md.

playwright-recording