generar-prd-conversacional
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No malicious instructions designed to bypass agent constraints or override system behavior were found in SKILL.md or the associated reference files.- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill does not access sensitive local files (e.g., .ssh, .env) or perform network operations to external domains.- [OBFUSCATION]: No base64, hex-encoded, or hidden unicode characters were detected in any of the analyzed files.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through conversational input but lacks dangerous capabilities.
- Ingestion points: User conversation input processed in SKILL.md workflow.
- Boundary markers: Uses
<!-- REVISAR: ... -->and<!-- PENDIENTE -->to delimit agent-generated content. - Capability inventory: Restricted to writing the PRD.md file.
- Sanitization: Relies on conversational validation steps before file writing.- [SAFE]: All external resource patterns and internal logic are consistent with the documented purpose of the skill.
Audit Metadata