The Agent Skills Directory

[COMMAND_EXECUTION]: The skill describes using standard search and analysis tools to audit codebases.
The instructions suggest using ripgrep (rg) for codebase searching and Semgrep or CodeQL for pattern analysis.
These tools are standard for the security auditing purpose of the skill and are used in a non-malicious context.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection due to its core function of processing untrusted codebases.
Ingestion points: The agent reads external codebase files using ripgrep, Semgrep, and CodeQL (SKILL.md).
Boundary markers: The skill does not specify explicit boundary markers or delimiters for the code content being analyzed.
Capability inventory: The skill utilizes codebase search tools and documentation generation capabilities.
Sanitization: There is no evidence of sanitization for the external code content before it is processed by the agent.

variant-analysis