objection-roleplay
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTIONNO_CODE
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses business-sensitive local files in the .sales/ directory including context, value propositions, and competitor positioning. While these are local operations, users should be aware that the agent accesses this data to personalize the roleplay.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by reading external context files and incorporating their content into roleplay prompts without sanitization or boundary markers. * Ingestion points: Reads .sales/context.md, .sales/value-proposition.md, and .sales/competitors.md. * Boundary markers: None. * Capability inventory: Reads files and appends results to .sales/objection-scripts.md. * Sanitization: None.
- [NO_CODE]: The skill consists only of markdown-based instructions and does not execute any scripts or download external dependencies.
Audit Metadata