The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill provides a run-code command that allows the execution of arbitrary JavaScript and Playwright code within the browser context to handle advanced automation tasks.
Evidence: The references/running-code.md file is entirely dedicated to the playwright-cli run-code command, providing examples for managing permissions, geolocation, and complex page interactions.
[DATA_EXFILTRATION]: The skill includes numerous commands to access, retrieve, and save sensitive browser data, including cookies, authentication tokens, and the entire browser state.
Evidence: SKILL.md and references/storage-state.md document commands like cookie-list, localstorage-get, sessionstorage-list, and state-save which can be used to export sensitive session information to the local file system.
[INDIRECT_PROMPT_INJECTION]: The skill automates interactions with external web content, creating a surface where malicious instructions on a webpage could influence the agent's behavior.
Ingestion points: Untrusted data enters the context through playwright-cli open, goto, and snapshot operations documented in SKILL.md.
Boundary markers: No specific delimiters or "ignore instructions" warnings are documented to prevent the agent from obeying instructions found within web content.
Capability inventory: The skill can perform file system writes (state-save, screenshot, pdf), network operations via the browser, and arbitrary code execution via run-code.
Sanitization: There is no evidence of content sanitization or validation for data retrieved from web pages before it is processed by the agent.

playwright-cli