The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The run-code command allows for the execution of arbitrary JavaScript within the browser environment. This provides a powerful dynamic execution surface that could be abused if the agent is directed to run untrusted code.
[DATA_EXFILTRATION]: Commands such as state-save, cookie-get, and localstorage-get provide access to sensitive browser session data. If redirected, this information (including authentication tokens) could be exposed or saved to files.
[COMMAND_EXECUTION]: The skill utilizes a set of playwright-cli commands to interact with the system and browser. The allowed-tools configuration scopes this to the playwright-cli binary.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8) because it processes content from external websites. * Ingestion points: playwright-cli snapshot, playwright-cli eval, and playwright-cli run-code results. * Boundary markers: None identified in the instruction set. * Capability inventory: Includes full browser navigation, interaction, file system writes (screenshots, PDFs, state files), and network request modification. * Sanitization: No explicit content sanitization or instruction filtering is documented for web page data before it is returned to the agent context.

playwright-cli