The Agent Skills Directory

[COMMAND_EXECUTION]: The management tools include the ability to send termination signals (SIGTERM) to processes that hold locks on the codebase index, as described in the codebase_stop tool in references/tool-reference.md.
[EXTERNAL_DOWNLOADS]: The skill facilitates the setup of indexing infrastructure by pulling Docker images and downloading embedding models from well-known technology providers such as Docker and Ollama.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and indexing external project files which could contain malicious instructions.
Ingestion points: codebase_index, codebase_update, and codebase_context_index (defined in references/tool-reference.md) read data from user-specified project paths.
Boundary markers: The instructions do not define specific delimiters or warnings to the agent to disregard instructions found within the indexed codebase files.
Capability inventory: The skill has the capability to read any file in the provided path, manage local processes, and communicate with external or local vector database and embedding services.
Sanitization: No sanitization or validation of the ingested file content is specified before the data is indexed.

codebase-management