The Agent Skills Directory

[SAFE]: The skill defines patterns for using browser automation tools provided by the chrome-devtools-mcp server. It does not contain malicious code, hidden instructions, or unauthorized network operations.
[DATA_EXPOSURE]: The skill instructs the agent to save session cookies and storage states to local files (e.g., /output/session_[domain].json) to maintain persistence and bypass repetitive logins. While this is a common automation practice, users should be aware that these files contain sensitive authentication tokens.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves processing data from external, untrusted websites, which creates a surface for indirect prompt injection.
Ingestion points: Untrusted web content is ingested via navigate_page and take_snapshot (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are provided for the content retrieved from tools.
Capability inventory: The skill possesses the capability to perform actions on the web including fill_form, upload_file, and navigate_page (SKILL.md).
Sanitization: No sanitization or validation of the ingested content is specified before the agent processes it.
[SAFE]: The skill explicitly advises against hardcoding credentials and recommends reading them from .env files, which is a standard and safe practice for secret management.

browser-pilot