The Agent Skills Directory

[SAFE]: The skill follows a controlled interactive workflow where every design item and final post must be explicitly approved by the user.- [SAFE]: All external interactions are conducted through standard, well-known development tools including the GitHub CLI (gh) and git.- [SAFE]: Local storage of design artifacts is limited to the project's .vapa workspace, maintaining a clear and isolated history of the design process.- [PROMPT_INJECTION]: The skill ingests untrusted data from GitHub issues (body and comments). While this constitutes an indirect prompt injection attack surface, the risk is mitigated by the mandatory user confirmation for all outputs and design decisions.\n
Ingestion points: GitHub issue content retrieved via gh issue view in Step 2.\n
Boundary markers: None specified in the instructions.\n
Capability inventory: Write for local files, gh issue comment for GitHub posts, and Bash for repository status.\n
Sanitization: None; relying on the interactive human-in-the-loop verification.

vapa-design