The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection (Category 8). It instructs the agent to ingest untrusted data from the user—specifically a 'component name' and 'purpose'—and use this information to scaffold a directory structure and create 10 files (e.g., projects/core/src/<component-name>/).
Ingestion points: The user provides the component name and purpose during the requirement gathering phase in SKILL.md.
Boundary markers: No specific boundary markers or 'ignore embedded instructions' warnings are provided for the user-supplied input.
Capability inventory: The skill possesses the capability to write multiple files to the filesystem and modify existing project files (e.g., projects/core/src/bundle.ts).
Sanitization: While the skill asks for 'kebab-case', it does not provide explicit validation or escaping instructions to prevent the agent from processing malicious input that could lead to path traversal or the injection of unexpected content into the boilerplate files.
[COMMAND_EXECUTION]: The skill instructs the agent to run local development commands (pnpm run lint and pnpm run test). These are legitimate actions within a development environment, but they involve subprocess execution based on the state of the newly created files.

component-creation