The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the rg (ripgrep) command-line tool to search for vulnerable code patterns within local directories (e.g., rg "PATTERN" targets/). This is a standard operation for static analysis and security auditing.
[EXTERNAL_DOWNLOADS]: The instructions direct the agent to query grep.app using constructed URLs (e.g., https://grep.app/search?q=PATTERN) to identify vulnerable patterns in public repositories. This involves outbound network requests to a well-known technology search service.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing and evaluating code matches from external sources (GitHub repositories via grep.app or local files). This introduces a surface for indirect prompt injection if the retrieved code snippets or comments contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Data enters the context through search results from grep.app or local repository searches as described in SKILL.md.
Boundary markers: Absent; the instructions do not specify the use of delimiters or 'ignore' instructions to encapsulate the content being audited.
Capability inventory: The skill leverages shell command execution (rg) and potentially browser-based search capabilities to inspect third-party source code.
Sanitization: There are no explicit sanitization or filtering steps defined for the external content before it is processed by the agent.

cross-pollination