platform-metadata-grounding-get

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves metadata such as labels, descriptions, and Apex class signatures from external Salesforce organizations, which represents a surface for indirect prompt injection if those fields contain malicious instructions.
  • Ingestion points: Data enters the context through the results of search_metadata, query_metadata, and describe_metadata calls to the Salesforce Catalog.
  • Boundary markers: The instructions do not prescribe specific character delimiters for tool output but do mandate that the agent distinguish between observed tool data and its own inferences.
  • Capability inventory: The skill's capabilities are restricted to read-only metadata exploration; it does not provide access to tools for writing files, executing arbitrary code, or initiating outbound network requests to third-party domains.
  • Sanitization: No specific sanitization or filtering logic is defined for the content retrieved from the org catalog.
  • [COMMAND_EXECUTION]: The skill references the use of the sf (Salesforce CLI) tool for metadata listing. This is a standard vendor-provided utility and its use is consistent with the skill's stated purpose of interacting with Salesforce orgs.
  • [SAFE]: All identified MCP servers and tools (metadata-grounding, headless-360) are official resources associated with the Salesforce platform, aligning with the author's identity.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 11:52 PM