deal-review

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill includes instructions to ignore system-level status alerts (such as 'not connected' or 'needs re-auth') in favor of attempting tool execution to improve reliability. It also restricts the agent from using standard discovery tools, prioritizing the paths defined within the skill.
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes data from Salesforce records, emails, Slack, and documents, which represents an ingestion surface for potentially untrusted content.
  • Ingestion points: Salesforce fields (Name, NextStep, Description), email threads, Slack messages, and documents.
  • Boundary markers: The skill does not define explicit delimiters to isolate external data from internal instructions.
  • Capability inventory: Capabilities include dispatch_readonly, display_widget, and sendMessage (used to draft communications).
  • Sanitization: No content-specific sanitization is specified for the data prior to analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 06:40 PM