The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill's primary function is the execution of Python and Shell commands for data processing and visualization. All examples demonstrate legitimate scientific use cases within a designated workspace.\n- [EXTERNAL_DOWNLOADS] (SAFE): While the skill mentions the availability of pip install for additional packages, this is a standard feature of a development environment and no specific malicious dependencies are referenced.\n- [PROMPT_INJECTION] (SAFE): The instructions do not contain attempts to override agent safety protocols, reveal system prompts, or bypass constraints. Instructional markers like 'CRITICAL' are used appropriately for operational guidance.\n- [DATA_EXFILTRATION] (SAFE): No hardcoded credentials, sensitive file path access (e.g., SSH keys), or suspicious network requests were found. Data output is constrained to the /workspace/output/ directory as per the UI design.

academic-python