The Agent Skills Directory

[SAFE]: The skill provides standard documentation for an in-process vector database library. All documented code snippets and usage patterns are legitimate and consistent with the library's stated purpose.
[EXTERNAL_DOWNLOADS]: The documentation notes that local embedding and reranking extensions download pre-trained models from public repositories like Hugging Face on first use, which is standard and expected behavior for these AI components.
[COMMAND_EXECUTION]: The skill includes standard installation instructions using pip and npm for the official library packages.
[PROMPT_INJECTION]: The database facilitates the ingestion of external data for indexing, representing a typical surface for indirect prompt injection if untrusted content is processed.
Ingestion points: Document ingestion via insert and upsert methods in references/data-operations.md.
Boundary markers: None are explicitly defined in the documentation.
Capability inventory: Supports file system writes via create_and_open and network-based embedding APIs.
Sanitization: The skill does not provide specific sanitization or validation routines for document content.

zvec