client-call-notes
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external and untrusted data (call transcripts, descriptions, or notes) that could contain hidden instructions intended to override the agent's behavior.
- Ingestion points: The skill relies on external "material" such as call transcripts or descriptions as its primary input source as defined in the "What to ask for" section.
- Boundary markers: The instructions lack specific delimiters or warnings to treat ingested text as data rather than instructions, potentially allowing embedded commands in a transcript to be executed.
- Capability inventory: The skill encourages the agent to interact with external tools, specifically suggesting the use of a "notes app" or the vendor's "Scratchpad connector" to save the processed output.
- Sanitization: There is no mention of sanitizing or validating the input data before it is formatted and saved to external systems.
Audit Metadata