The Agent Skills Directory

[SAFE]: The skill performs legitimate documentation tasks using authorized file system tools (Read, Write, Edit, Glob, Grep). No suspicious command execution or persistence mechanisms were detected.
[PROMPT_INJECTION]: The skill ingests external data from specification files, creating an indirect prompt injection surface. However, the potential impact is significantly mitigated by the lack of high-risk capabilities like shell access or network connectivity. Evidence: 1. Ingestion points: Specification file content (SKILL.md workflow step 1); 2. Boundary markers: Employs standard Markdown headers and comment blocks as delimiters; 3. Capability inventory: Read, Glob, Grep, Write, Edit, Skill, Task; 4. Sanitization: None documented for input specification content.

adr-create