The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill uses a user-supplied project name as a template variable {project} that is interpolated into file paths (e.g., src/{project}/), Docker configurations, and Python import statements.
Ingestion points: The user-provided project name is captured in SKILL.md via the instruction "Ask the user for the project name if not provided".
Boundary markers: There are no explicit validation rules or boundary markers defined to constrain the project name string.
Capability inventory: The skill instructs the agent to perform extensive file-write operations and generates a justfile and Dockerfile containing shell commands.
Sanitization: No sanitization or validation is performed on the user-provided input, which could allow a malicious user to attempt path traversal or command injection via crafted project names during the scaffolding process.

fastapi-setup