The Agent Skills Directory

[SAFE]: The skill serves as a technical debugging guide, instructing the agent to gather diagnostic data (e.g., clipboard contents, HTTP headers, DOM structures) before proposing fixes. It includes case studies and templates for common debugging scenarios.
[DATA_EXPOSURE]: The methodology encourages the agent to request potentially sensitive diagnostic information from the user, such as raw clipboard data via navigator.clipboard.read() or full HTTP responses via curl -v. While appropriate for the intended purpose of debugging, users should be aware that this data may contain credentials or PII if not sanitized.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to handle data from untrusted external sources (browsers, third-party APIs). This presents a surface for indirect prompt injection if the ingested data contains malicious instructions. The skill partially mitigates related risks by explicitly advising against using innerHTML and suggesting safer DOM manipulation methods like TreeWalker to prevent XSS.

cc-external-system-debugging