The Agent Skills Directory

[SAFE]: No malicious patterns such as data exfiltration, obfuscation, or persistence were detected. The skill follows standard developer workflows for browser automation and video processing.
[COMMAND_EXECUTION]: The skill generates TypeScript automation scripts and executes them locally using npx tsx. It also utilizes system utilities like ffmpeg, ffprobe, and the macOS say command for audio generation.
[EXTERNAL_DOWNLOADS]: The skill performs network checks on the local application URL using curl and downloads Playwright browser binaries via npx playwright install.
[PROMPT_INJECTION]: The skill processes untrusted user input (the 'Task' and 'Selectors' fields) to generate executable code. While this constitutes an indirect prompt injection surface, the risk is mitigated by mandatory manual review gates (Gate A, Gate C, and Gate D). (1) Ingestion points: User-supplied task descriptions and CSS selectors enter the context at activation. (2) Boundary markers: Absent in the code generation templates. (3) Capability inventory: Includes file system writes and local command execution via npx and system shells. (4) Sanitization: User input is interpolated directly into script templates without explicit sanitization or escaping.

playwright-tutorial-video