service-catalog-template-search
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes template names and descriptions from the Salesforce Unified Catalog API. This external data represents a potential attack surface where malicious content within a template's metadata could attempt to influence the agent's ranking or recommendations.\n
- Ingestion points: The skill fetches the catalog using the Salesforce CLI (
sf) and stores the output in a temporary file (/tmp/uc-search-get.txt) before processing.\n - Boundary markers: The skill contains explicit constraints for the agent to treat template text strictly as untrusted data and never as instructions, ensuring the agent remains in a data-processing role.\n
- Capability inventory: The skill employs shell execution for both the Salesforce CLI and a local Node.js script to perform data retrieval and parsing.\n
- Sanitization: A deterministic classification script (
scripts/classify-catalog.mjs) is used to parse the API response, ensuring the agent interacts with structured, validated data rather than raw, potentially manipulative text output.
Audit Metadata