The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions direct the agent to consume project-specific context from files such as .agents/product-marketing.md and .claude/product-marketing.md. This ingestion of external data constitutes an indirect prompt injection surface. The skill does not provide boundary markers or sanitization logic to prevent the agent from executing instructions potentially hidden within these context files. Findings: 1) Ingestion points: .agents/product-marketing.md, .claude/product-marketing.md, product-marketing-context.md (as specified in SKILL.md); 2) Boundary markers: Absent; 3) Capability inventory: The skill contains no executable scripts or system-modifying tools; 4) Sanitization: Absent.
[EXTERNAL_DOWNLOADS]: The skill references an external tool registry and integration guides hosted at https://github.com/mbzadev/skills. These links point to a repository that uses a naming variation of the author name. While these are informational resources, they represent a dependency on content not contained within the skill package.

revops