The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to gather context about the project's current state. Evidence: In SKILL.md, Phase 1 of the workflow uses 'git diff --name-only HEAD~1' to infer targets for review based on recent changes.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by processing untrusted data from the local repository to drive its advisory logic. * Ingestion points: The skill reads various local project files in SKILL.md, including implementation code, design tokens (tokens.json), and framework configurations (tailwind.config.*). * Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions potentially embedded within the analyzed source code or configuration files. * Capability inventory: The skill has the ability to read arbitrary files from the filesystem, execute git commands, and invoke other functional skills such as charting, animations, and storybook. * Sanitization: The content retrieved from the files is used directly to inform design decisions and generate implementation code without validation or sanitization of embedded text.

visual-design