ccf-visual-composer

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes a bundled Python library (ccfa_plot_recipes.py) designed to generate editable SVG plots using only standard library modules. The instructions direct the agent to adapt these recipes for specific data visualization tasks, which is a standard and well-controlled use of the execution environment for scientific deliverables.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied manuscripts, architectural notes, and data to generate specifications and prompts for image generation tools.
  • Ingestion points: Technical content and architectural descriptions are extracted from the user's project files as described in references/architecture-diagram-generation.md.
  • Boundary markers: The skill uses a 'Visual Contract' (references/visual-contract.md) and a 'Diagram Specification' to structure and delimit external data.
  • Capability inventory: The skill utilizes Python execution for plotting and the host's image generation capability (GPT Image 2).
  • Sanitization: The skill mitigates the risk of processing malicious data by requiring a 'Required GPT Image 2 Confirmation Gate,' where the agent must display the derived prompt and obtain explicit user authorization before invoking the external generation tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 08:26 AM
Security Audit — agent-trust-hub — ccf-visual-composer