The Agent Skills Directory

[COMMAND_EXECUTION]: Provides templates for common project initialization commands using standard development tools like npm, uv, go, and cargo.
[CREDENTIALS_UNSAFE]: Correctly implements secret management patterns by using .env.example files and validation libraries. No hardcoded sensitive data is present; all credentials in templates are non-functional placeholders.
[EXTERNAL_DOWNLOADS]: References official images and actions from trusted organizations and well-known services (such as Docker Hub, GitHub, and Astral-sh) in its scaffolding templates.
[PROMPT_INJECTION]: Potential surface for indirect prompt injection exists because the skill generates project files based on user-provided specifications.
Ingestion points: User project requirements and names (SKILL.md)
Boundary markers: Absent
Capability inventory: Write, Edit, and Bash tools (SKILL.md)
Sanitization: Absent

scaffold