churn-prevention
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to read local files such as
.agents/product-marketing.mdto gather product context. This represents an ingestion point for untrusted data that could lead to indirect prompt injection if those files are compromised. - Ingestion points:
.agents/product-marketing.md,.claude/product-marketing.md, andproduct-marketing-context.md(SKILL.md). - Boundary markers: Absent; instructions do not specify the use of delimiters for this data.
- Capability inventory: The skill is primarily used for generating strategic advice and includes no subprocess calls, file-write, or network operations in its own instructions.
- Sanitization: Absent; no mention of sanitizing or escaping the file content.
- [EXTERNAL_DOWNLOADS]: The skill contains references to documentation and tool registries hosted on the author's GitHub repository (
github.com/coreyhaines31/marketingskills) and official developer documentation for the Paddle payment platform. These are documented as vendor-specific resources and well-known services.
Audit Metadata