The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses rg (ripgrep) and grep to perform keyword searches on local files within the workspace. These tools are used appropriately for searching project documentation and do not involve unvalidated shell interpolation.
[INDIRECT_PROMPT_INJECTION]: The skill functions as a retrieval mechanism that fetches content from external Pinecone indexes and local markdown files. This ingestion of external data creates a surface for indirect prompt injection if the retrieved snippets contain malicious instructions intended to influence the calling agent.
Ingestion points: Local markdown files in the working-folder/ and result metadata from the Pinecone index.
Boundary markers: The skill returns raw text snippets and citations; it does not implement specific delimiters or instruction-ignore warnings in its return format.
Capability inventory: Accesses files via Read and executes search commands via Bash (grep/rg).
Sanitization: The skill returns raw content without sanitizing or escaping the retrieved text, delegating processing logic to the calling skill.

semantic-recall