vector-db-setup
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements logic for ingesting and processing untrusted external document content to be stored in vector databases for semantic search.
- Ingestion points: The
ingestDocumentsfunction inSKILL.mdaccepts raw document content for chunking and embedding. - Boundary markers: The code snippets do not utilize explicit boundary markers or instructions to the model to ignore potential commands embedded within the processed text chunks.
- Capability inventory: The skill contains implementation for database write operations (upsert/add) and similarity search queries (query/search) across multiple database providers.
- Sanitization: There is no provided logic for sanitizing or filtering external content before it is processed by embedding models or indexed in the databases.
Audit Metadata