The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides curl command templates in references/wayback-machine.md for querying the Wayback Machine API. These are standard GET requests used for archive discovery and status checks.
[EXTERNAL_DOWNLOADS]: The skill fetches archived data and web page replays from well-known services hosted at archive.org and web.archive.org.
[DYNAMIC_EXECUTION]: The skill utilizes evaluate_script within SKILL.md to execute JavaScript in the browser context. This is used to extract information that is not available in the standard accessibility tree.
[INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection because it processes untrusted data from external websites.
Ingestion points: Web content is ingested into the agent context through take_snapshot and evaluate_script in SKILL.md.
Boundary markers: Absent. The instructions do not provide explicit delimiters or ignore-instructions for the data retrieved from web pages.
Capability inventory: The skill can execute browser-side scripts (SKILL.md), write screenshots and data to the local filesystem (SKILL.md), and perform network operations via curl (references/wayback-machine.md).
Sanitization: Absent. The skill does not define methods for sanitizing or validating web-based content before it is processed by the AI.

chromium-browser