embedding-optimization
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [SAFE]: The skill implements standard vector embedding optimization techniques using industry-standard libraries and follows secure development practices.
- [EXTERNAL_DOWNLOADS]: Fetches pre-trained models from Hugging Face's public model repository via the sentence-transformers library.
- [DATA_EXFILTRATION]: Accesses external embedding APIs (OpenAI, Cohere, Voyage AI) using credentials supplied through environment variables, following standard secret management practices.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests text data from JSONL files to generate embeddings. While this provides a surface for untrusted data, the processing is a passive data operation, and standard downstream safety protocols apply.
Audit Metadata