define-character

Warn

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the magick utility to process images using variables derived from repository files.- Evidence: The Recompose assembly section in SKILL.md contains shell commands that interpolate variables such as <character>, <left padding>, <cell width>, and <appearance> into magick calls.- Risk: If the source files (docs/game/assets.md or docs/game/world.md) are modified by an attacker to include shell metacharacters, it could lead to arbitrary command execution.- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection.- Ingestion points: docs/game/world.md, docs/game/assets.md, docs/game/story.md, and docs/game/screenplay.md (referenced in SKILL.md).- Boundary markers: Absent. The skill instructions direct the agent to quote entire sections like ## Visual direction directly into generation prompts.- Capability inventory: The agent has the ability to execute shell commands (magick) and write to the source code directory (src/).- Sanitization: No sanitization or validation of the ingested markdown content is performed before processing.- [EXTERNAL_DOWNLOADS]: The skill relies on external binary tools and project-level dependencies.- Evidence: The instructions require ImageMagick 7 to be installed and reference documentation and recipes within the fondale package under node_modules.- Note: While these are standard requirements for the stated purpose, they represent a dependency on external code and environment configuration.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 21, 2026, 05:00 PM