The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local shell commands such as grep to scan Python files for specific library imports (e.g., openai, torch, sklearn) to determine the project type. It also uses the mlflow CLI and jq to retrieve and parse experiment metadata stored in temporary local files.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes local codebase content to categorize the project. This creates a theoretical surface for indirect prompt injection if project files contain malicious text patterns designed to manipulate the agent's logic. However, the risk is minimal as the analysis is limited to identifying specific import statements and code patterns.
Ingestion points: Local project files scanned via grep and experimental tags retrieved via CLI. (SKILL.md)
Boundary markers: Not specified.
Capability inventory: Local file system read access (grep), execution of local CLI tools (mlflow, jq), and suggested modification of application initialization code.
Sanitization: Not present.
[EXTERNAL_DOWNLOADS]: Directs users to several quickstart guides and tutorials hosted on mlflow.org. As these are official resources owned by the skill's authoring organization, they are considered safe references.

mlflow-onboarding