find-internal-answer

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill operates by retrieving and processing content from external sources including Slack threads, emails, and internal documentation, which constitutes an attack surface for indirect prompt injection.
  • Ingestion points: User-supplied questions and content retrieved from Slack, email, and document search results (SKILL.md).
  • Boundary markers: Absent. The instructions do not define specific delimiters or warnings to prevent the agent from following instructions potentially embedded within the retrieved data.
  • Capability inventory: The skill utilizes dispatch_readonly and dispatch tools for Salesforce access, alongside generic search and read capabilities for communication and document platforms.
  • Sanitization: Absent. No mention is made of filtering, escaping, or validating the content retrieved from these sources before it is interpreted by the agent.
  • [COMMAND_EXECUTION]: The skill provides a specific, hardcoded tool call to verify Salesforce connectivity using a standard GraphQL identity query.
  • Evidence: The skill instructs the agent to run dispatch_readonly(method: "GET", url: "/services/data/v66.0/graphql", queryParams: { "queryInput": "{\"query\":\"query { uiapi { currentUser { Id } } }\"}" }) to confirm server access regardless of reported connection status. This is a standard, low-risk diagnostic operation.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 06:39 PM