The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied strings such as the app name, description, and author, and interpolates them directly into generated Python, TOML, and Markdown files.
Ingestion points: User requirements gathered during the initialization step in SKILL.md.
Boundary markers: No explicit boundary markers or instructions to ignore embedded content are included in the generated boilerplate code.
Capability inventory: The skill performs file system write operations and directory creation based on user-provided metadata.
Sanitization: The skill enforces snake_case for names, but does not perform escaping or sanitization on free-text fields like the description or author strings.
[COMMAND_EXECUTION]: The skill executes diagnostic commands to verify the current working directory and environment state.
Execution method: Shell execution of pwd and ls -la via agent tools.
Scope: Limited to verifying that the user is within a valid Frappe Bench apps directory.

frappe-app