The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates searching and reading from the codebase and PDF documents, creating a surface for indirect prompt injection where malicious instructions in files could influence the agent.
Ingestion points: Content retrieved from the codebase via exec:codesearch and Read (SKILL.md).
Boundary markers: Absent; there are no instructions to use delimiters or ignore instructions within search results.
Capability inventory: The skill has access to exec:codesearch, Read, and exec:nodejs (SKILL.md).
Sanitization: Absent; the skill does not specify any validation or filtering of the content read from files.
[COMMAND_EXECUTION]: The instructions suggest using exec:nodejs to execute fs.readdirSync for directory listings. This allows for dynamic execution of JavaScript code to interact with the local filesystem.

code-search