The Agent Skills Directory

[SAFE]: The skill is composed entirely of natural language instructions and structural templates for design feedback management. It does not perform any network operations, system-level commands, or external code execution.- [DATA_EXPOSURE]: While the skill references specific file paths such as _workspace/feedback/round-{N}.md, concept.md, and slots.json, these are internal project artifacts consistent with the skill's stated purpose of font design management. No sensitive system paths (e.g., SSH keys, credentials) are accessed.- [INDIRECT_PROMPT_INJECTION]: The skill handles untrusted user input (design feedback) but includes safety-aligned practices by instructing the agent to wrap user feedback in blockquotes (>) for provenance and to generalize feedback into technical specifications rather than executing it as instructions.

design-feedback