version: "1.0.0" schema_version: 2 name: image-quality-and-artifact-analyzer kind: tool description: Analyzes a single edited image for visual artifacts, distortions, blurriness, unnatural lighting, and over-editing. input_schema: images: list[base64_str] query: str output_schema: flaws: list[str] quality_score: int blending_quality: str system_prompt: You are an expert image quality evaluator. Analyze the provided edited image (compared to the source) for visual artifacts, distortions, blurriness, unnatural lighting, over-editing, and poor blending of edited elements. Specifically check how well new objects are integrated into the scene in terms of lighting, shadows, and perspective. Return a JSON with keys "flaws" (list of specific issues found), "quality_score" (1-4 integer), and "blending_quality" (description of how well edits blend with the original).

Image Quality and Artifact Analyzer

Use this tool to objectively assess the visual quality of the edited image compared to the source.

When to use:

• When you need to assess visual quality objectively before scoring.
• When you suspect the edited image has artifacts, distortions, or unnatural blending.

Example Call:

<tool>{"name": "image-quality-and-artifact-analyzer", "images": ["source", "edited"], "query": "Analyze the edited image for any artifacts, distortions, or unnatural lighting introduced by the edit."}</tool>

<!-- Release review: RewardHarness v0.4.1 (2026-08-26). -->