gan-style-harness

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues detected. The skill provides architectural guidance and documentation for a multi-agent workflow involving the coordination of multiple AI agents (Planner, Generator, Evaluator).
  • [PROMPT_INJECTION]: The skill describes a workflow that processes data from generated files (spec.md, feedback-NNN.md) using powerful system tools (Bash, Write, Playwright). While inherent to development agents and necessary for the stated purpose, this creates a surface for indirect prompt injection if external or malicious data enters these intermediate files.
  • Ingestion points: spec.md, feedback-NNN.md (via Read and Bash tools).
  • Boundary markers: None specified in the provided prompts/instructions.
  • Capability inventory: Read, Write, Edit, Bash, Grep, Glob, Task, Playwright MCP (documented in SKILL.md).
  • Sanitization: No explicit sanitization or validation of the generated spec or feedback files is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 10:25 PM