deal-signals
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple external communication channels to identify deal risks and evidence.
- Ingestion points: In step 2b (Evidence), the skill specifically instructs the agent to ingest content from email (replies, bounces, departures), documents/transcripts, and Slack messages.
- Boundary markers: The instructions do not define boundary markers or delimiters for this external data, nor do they include explicit instructions for the agent to ignore command-like patterns within that data.
- Capability inventory: The agent has access to the
dispatchtool (Salesforce write operations) and provides adisplay_widgetwith asendMessageaction, which could be triggered by instructions embedded in the external data. - Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from external sources before it is interpreted by the agent or included in the final output.
Audit Metadata