The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides tools for executing arbitrary JavaScript code within the browser context via playwright-cli eval and playwright-cli run-code. This allows for dynamic execution based on agent-generated or potentially external inputs.
[EXTERNAL_DOWNLOADS]: The instructions recommend using npx playwright-cli, which facilitates the download and execution of the package from the public npm registry at runtime if it is not already present.
[DATA_EXFILTRATION]: The skill includes extensive functionality for accessing and extracting sensitive browser data, including cookies (cookie-get), LocalStorage, and SessionStorage. It also allows saving full authentication states (including session tokens) to local files using state-save.
[PROMPT_INJECTION]: The skill is highly vulnerable to indirect prompt injection.
Ingestion points: External websites via goto and open commands in SKILL.md.
Boundary markers: The snapshot mechanism described in the 'Snapshots' section does not specify delimiters or warnings to ignore instructions found within page content.
Capability inventory: The skill has significant capabilities, including file system writes (state-save, screenshot), sensitive data reading (cookie-get), and arbitrary code execution (eval).
Sanitization: There is no evidence of sanitization or filtering for the data extracted from web snapshots before it is presented to the agent.

playwright-cli