paper-figure

Warn

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill dynamically generates Python scripts (gen_fig*.py) containing plotting logic and data processing instructions.
  • Evidence: Step 4 provides templates for generating standalone Python scripts for various figure types.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute the dynamically generated Python scripts.
  • Evidence: Step 5 includes a shell loop that executes all generated scripts in the environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests data from external files (PAPER_PLAN.md, experiment logs) and incorporates that data into both executable scripts and LLM prompts.
  • Ingestion points: Reads figure plans from PAPER_PLAN.md and numerical data from JSON/CSV files.
  • Boundary markers: No delimiters or sanitization logic is specified for separating data from instructions in the generated scripts or the LLM prompts.
  • Capability inventory: Uses Bash(*) for script execution and mcp__codex__codex for model review.
  • Sanitization: None described for external content before interpolation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 22, 2026, 06:20 PM