The Agent Skills Directory

[SAFE]: The skill operates as a structured interview process to generate markdown and JSON documentation. It follows a "propose → confirm → lock" workflow which ensures human oversight of all generated content.
[SAFE]: The security section of the rubric (Section 06) promotes best practices by requiring explicit definitions for secrets handling, access models, and data classification, specifically identifying "what is never logged" to prevent accidental data exposure.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it reads from upstream files such as VISION.md, business-model, and operations. However, these are internal workspace files rather than untrusted external sources. The skill lacks sensitive capabilities (like arbitrary network access or high-privilege command execution) that would make this ingestion surface a high-risk vector.

technology