signup-flow-cro
Pass
Audited by Gen Agent Trust Hub on Feb 22, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [Prompt Injection] (SAFE): The instructions use standard persona-based roles without any attempt to bypass security filters or override system constraints.
- [Data Exposure] (SAFE): The skill mentions reading a local file (.claude/product-marketing-context.md) to gain context, which is a standard procedure for personalized assistants and does not target sensitive system credentials.
- [Remote Code Execution] (SAFE): There are no scripts, command-line instructions, or remote execution patterns present.
- [Indirect Prompt Injection] (SAFE): While the skill defines a surface for ingesting external context, it does not possess any exploitable capabilities like network access or shell execution that could be triggered by malicious data.
Audit Metadata