The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection due to its core functionality of processing external data.
Ingestion points: Untrusted data enters the agent's context in Step 1 (fetching the user-provided audit URL) and Step 3 (fetching the top 10 Google search results for the primary keyword).
Boundary markers: The instructions lack explicit delimiters (e.g., XML tags or backticks) or specific 'ignore' instructions to isolate the fetched external content from the agent's internal reasoning logic.
Capability inventory: The agent possesses fetch and google capabilities. While these are used for research, an injection in the fetched content could attempt to manipulate the audit output or lead the agent to visit malicious URLs.
Sanitization: The skill does not implement sanitization or validation of the fetched content to ensure it does not contain executable instructions or malicious prompts.

page-audit