Gemini MCP Server for APICore - 基于 APICore 渠道的 Gemini 2.5 Flash Image 图片处理服务
Single tool 'generate_image' has a complete JSON Schema with typed parameters and descriptions. However, the tool description is in Chinese and lacks clarity on when to use it vs alternatives. Parameter descriptions are present but verbose (mixing examples with constraints). No output schema documented. Error handling returns generic text responses without recovery guidance. Tool naming is clear (verb_noun), but descriptions could be more concise and LLM-optimized.
使用 Gemini 2.5 Flash Image (APICore) 生成图片,支持文生图和图生图两种模式,又被称为 nano banana
Output schema not documented. Tool returns unstructured text response without specifying fields like image_url, file_path, dimensions, or generation_time. LLMs cannot plan downstream operations or extract structured data.
Error handling returns generic text ('❌ 错误: <message>') without recovery guidance. LLM cannot determine if error is retryable, user-fixable, or fatal. No actionable next steps provided.
Parameter descriptions mix examples with constraints (e.g., '例如:"一只可爱的小猫咪"'). LLMs tend to reuse example values literally. Use enums or format constraints instead of inline examples.
Tool description is in Chinese and lacks English translation. Descriptions should be in the primary language of the MCP ecosystem (English) for discoverability and LLM comprehension.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
No input validation or constraint documentation for 'output_dir' parameter. Path traversal attacks possible if untrusted input provided. No mention of allowed directory patterns or security boundaries.