The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local shell commands such as ls and bench to query the project's state. These operations are restricted to the local environment and are essential for the skill's purpose of identifying project metadata.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes external data including directory names, hooks.py, and common_site_config.json to identify the project type. While this creates a surface for indirect injection, the risk is negligible for a discovery-focused triage tool.
Ingestion points: File system structure, project configuration files, and command outputs from the local environment.
Boundary markers: None identified.
Capability inventory: Shell command execution (bench, ls) and interactive Python console access via bench console.
Sanitization: Not implemented, as the skill focuses on data discovery rather than processing untrusted user input.
[DYNAMIC_EXECUTION]: The skill instructs the agent to use bench console to execute specific Python commands (e.g., frappe.__version__). This is a standard method for interacting with the Frappe framework and does not involve untrusted remote code or unsafe deserialization.

project-triage