The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/build_result_contracts.py generates new Python executable files (result_contract_io.py, run_modeling.py, and q*_model.py) by writing pre-defined code templates to the paper_output/code/modeling/ directory.
[COMMAND_EXECUTION]: The run_modeling.py script generated by this skill contains logic to iterate through and execute Python scripts (q*_model.py) using subprocess.run.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Reads metadata and configuration from paper_output/plan/model_route.json, data_plan.json, and visualization_plan.json.
Boundary markers: Absent; the skill does not use specific delimiters to separate untrusted data from instructions.
Capability inventory: Includes file writing (Path.write_text) and subprocess execution (subprocess.run) in the generated scripts.
Sanitization: Uses safe_slug for filename generation and json.dumps for data interpolation, which reduces the risk of code injection if the input source is compromised.
Analysis: The skill facilitates a workflow where external configuration data is transformed into executable code, creating a surface where malicious input could influence the generated script behavior.

model-code-and-result-generator