azure-search-documents-ts
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS] (SAFE): The skill references official Microsoft Azure packages (@azure/search-documents, @azure/identity). These are trusted official dependencies and do not trigger remote code execution concerns.
- [CREDENTIALS_UNSAFE] (SAFE): Documentation uses placeholders for sensitive information (e.g., '') and explicitly warns against hardcoded credentials in the anti-patterns section, recommending DefaultAzureCredential.
- [DATA_EXFILTRATION] (SAFE): Network communication is limited to the user-defined Azure Search endpoint. No evidence of unauthorized data transfer to third-party domains.
- [PROMPT_INJECTION] (SAFE): No injection patterns or instructions to override agent behavior were found in the text or code snippets.
- [INDIRECT_PROMPT_INJECTION] (LOW):
- Ingestion points: searchClient.search() and searchClient.getDocument() in SKILL.md.
- Boundary markers: Standard LLM handling; documentation encourages structured query patterns.
- Capability inventory: uploadDocuments, deleteDocuments, and createOrUpdateIndex in SKILL.md.
- Sanitization: Documentation promotes the 'odata' tagged template helper in references/acceptance-criteria.md for safe filter construction.
Audit Metadata