The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/test_estimator.py utilizes subprocess.run to automate the testing of the run_estimator.py command-line interface. This execution is limited to the skill's own local Python files using the system's own interpreter (sys.executable) and does not involve shell interpolation or untrusted input, making it a standard and safe development practice.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided biological target names which are subsequently passed to specialized research tools and included in reports. These inputs are treated as parameters for domain-specific queries and are not used in contexts where they could influence the agent's core logic or bypass safety guardrails.
[SAFE]: All external dependencies (e.g., pandas, python-docx, requests) are well-known, standard libraries for data processing and report generation. The skill's workflow is transparent and focused entirely on its stated purpose of industrial costing and resource planning.

experiment-dataset-costing