eval-harness
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements standard development workflows for software testing and evaluation. It uses shell commands for deterministic grading and model prompts for qualitative evaluation.
- [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection by processing project files and tool outputs to inform the model's grading behavior.
- Ingestion points: Reads evaluation definitions from .gemini/evals/*.md and output from shell tools.
- Boundary markers: The skill uses structured Markdown identifiers like [CAPABILITY EVAL] and [MODEL GRADER PROMPT] to organize external content.
- Capability inventory: The skill uses Read, Write, Edit, and Bash tools to manage the evaluation lifecycle.
- Sanitization: There are no explicit instructions for sanitizing or escaping content from external files before processing.
Audit Metadata