forecast-narrative
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 Salesforce records which could contain malicious instructions.
- Ingestion points: Salesforce Opportunity record fields (e.g.,
Name,NextStep,StageName) and field history values are ingested viadispatch_readonlycalls inSKILL.md. - Boundary markers: The instructions lack explicit delimiters or warnings to ignore embedded instructions within the queried data (e.g., in the
NextSteporNamefields). - Capability inventory: The skill has the capability to read Salesforce data and render user-facing widgets (
display_widget). It does not have access to shell execution or arbitrary file-system writes. - Sanitization: No sanitization or validation of the ingested record content is specified before it is used to populate the
display_widgetor the text fallback narrative.
Audit Metadata