Loading ToolBench
Gemini 3.1 Flash Image MCP server - fast image generation with advanced reasoning, 512px-4K resolution, up to 14 reference images, Google Search grounding, and configurable thinking mode
Three tools with complete schemas and descriptions, but significant gaps in parameter documentation and error handling. generate_image has 12 parameters with descriptions, but many lack type constraints (e.g., aspect_ratio, image_size, output_format should be enums). batch_generate duplicates generate_image's parameters without clear composition guidance. generate_app_icon is well-scoped but lacks output schema documentation. No tool describes error recovery paths or categorizes failures. Naming is clear (verb_noun pattern), but parameter relationships are underdocumented.
Generate multiple images in parallel using batch processing. Accepts a list of prompts and generates them concurrently with shared parameters.
Purpose-built app icon and logo generator. Automatically enforces square (1:1), transparent, 1024px PNG output. Prompt must describe only the subject/artwork, not the deliverable format.
Generate images using Gemini 3.1 Flash Image with support for multiple aspect ratios, resolutions (512px-4K), reference images (up to 14), Google Search grounding, and configurable thinking mode.
Parameters lack enum constraints. aspect_ratio, image_size, output_format, thinking_level, response_modalities accept free-form strings instead of enums. LLMs will hallucinate invalid values like '2:1' or '3K'.
No output schema documented. Tools return images but do not specify response structure (file path, URL, metadata, dimensions, format). LLMs cannot plan downstream operations or extract results reliably.
No error handling guidance. Tools do not document failure modes (API quota exceeded, invalid reference image, network timeout) or recovery steps. LLMs cannot self-correct or retry intelligently.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | <=2025-11-25 | v2 |
batch_generate duplicates generate_image parameters without composition clarity. No guidance on when to use batch vs. sequential calls, or how batch results map to individual prompts. Violates single-responsibility principle.
Parameter relationships undocumented. reference_image_paths accepts 'string | list[str] | null' but no guidance on format (file path, URL, base64). enable_google_search and enable_image_search are separate flags with no explanation of interaction or precedence.