figure-bible

Warn

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DYNAMIC_EXECUTION]: The scripts scripts/init_theme.py and scripts/validate_theme.py perform dynamic path manipulation to include a shared library directory in the Python search path. Evidence includes the use of sys.path.insert(0, str(Path(__file__).resolve().parents[3] / "lib")) to resolve the library location at runtime.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text through command-line arguments which are subsequently stored in a configuration file used to generate AI prompts.
  • Ingestion points: The scripts/init_theme.py script accepts values for --style, --negative, --codex-model, and --codex-effort via CLI.
  • Boundary markers: The provided implementation does not include specific delimiters or warnings to prevent the downstream interpretation of user-provided content as commands.
  • Capability inventory: The skill has the capability to write the resulting configuration to the filesystem via args.out.write_text in scripts/init_theme.py.
  • Sanitization: Input validation is limited to hex color checks; other textual inputs are trimmed but not sanitized against potential prompt injection attacks.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes the well-known jsonschema package for theme validation, which is installed via the package manager during execution.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 21, 2026, 09:05 AM