The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest a DESIGN.md file from the local project root to guide its decision-making process for UI development and architectural auditing. This creates a surface where a malicious file in a repository could influence the agent's output or behavior.
Ingestion points: The agent is instructed to read DESIGN.md and source code files from the project directory.
Capability inventory: The skill allows the agent to write new component, page, and store files, and to execute shell commands for package installation.
Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are provided; the agent is told to "follow it strictly."
Sanitization: There is no evidence of validation or filtering for the content within the DESIGN.md file.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands to manage UI components using a well-known developer tool.
Evidence: Use of pnpm dlx shadcn@latest add <component> for UI primitive installation in SKILL.md.

react-rules