account-context

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from multiple external sources (Salesforce records, emails, documents, and Slack messages), which represents an indirect prompt injection attack surface.
  • Ingestion points: Salesforce GraphQL responses, email threads, document content, and Slack message history are retrieved via MCP tools and integrated into the agent's context in SKILL.md.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard instructions that might be embedded within the retrieved third-party data.
  • Capability inventory: The skill possesses read/write capabilities within Salesforce (dispatch, dispatch_readonly) and can render UI elements through display_widget.
  • Sanitization: While the skill emphasizes "grounding" and "citing every value exactly," it does not explicitly define sanitization or filtering logic for the content retrieved from external sources before synthesis.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 06:39 PM