data-cloud-2025
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents patterns for ingesting untrusted data from external sources (e.g., S3, SFTP, and web transcripts) into the Salesforce Data Cloud and its Vector Database (detailed in references/ingestion-patterns.md and references/vector-database.md). This establishes a potential surface where external content could influence downstream agent behavior.
- Ingestion points: External data stream ingestion and unstructured document indexing for vector search.
- Boundary markers: None present in the provided architectural examples.
- Capability inventory: The snippets demonstrate database writes and network callouts via Apex.
- Sanitization: Reference code focus is on integration logic and does not implement input validation or content filtering.
Audit Metadata