The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/validate-design-system.sh contains an inline Python script that is executed by piping a heredoc into the python3 interpreter. This script processes analysis findings and performs an atomic file write to .storybook/design-system-health.json.
[INDIRECT_PROMPT_INJECTION]: The skill processes DESIGN.md, which is explicitly described in SKILL.md and scripts/validate-design-system.sh as untrusted and a claim that can drift or lie. It also emits a prompt for an LLM sub-agent to perform semantic checks on naming drift and style conflicts across project files. Ingestion points: DESIGN.md and project source files (CSS/TSX). Boundary markers: None. Capability inventory: File system write access and shell command execution via the Bash and Write tools. Sanitization: Uses standard text processing tools like grep and awk.
[COMMAND_EXECUTION]: The bash scripts execute various system commands and developer tools, including npx stylelint, grep, find, and python3, to perform scanning and file scaffolding within the .storybook/ directory.

sb-health