The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill generates Python scripts at runtime (e.g., paper_plot_style.py and gen_fig*.py) and executes them to produce figures. This involves assembling executable code from internal templates and project data, which is a significant capability.
[COMMAND_EXECUTION]: The skill uses the bash tool to execute the generated Python scripts in a loop (for script in gen_fig*.py; do python "$script"; done).
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the project environment which is used to influence code generation and external model queries.
Ingestion points: The skill reads and parses PAPER_PLAN.md and experiment data files such as exp.json or CSV files.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands when the agent processes the contents of these files.
Capability inventory: The skill can write files, execute arbitrary Python code via bash, and call the mcp__codex__codex tool to interact with other LLM models.
Sanitization: The skill lacks sanitization or validation of the data ingested from PAPER_PLAN.md or experiment logs before it is interpolated into Python scripts or sent to the REVIEWER_MODEL for evaluation.

paper-figure