zhenzhen-nanobanana-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| NANOBANANA_OUTPUT_DIR | No | Optional absolute path for generated image output. Defaults to the project's outputs/ directory. | |
| ZHENZHEN_NANOBANANA_API_KEY | Yes | API key for the t8star image service. Use the raw key without a 'Bearer ' prefix. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": true
} |
| resources | {
"subscribe": false,
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| list_modelsA | 列出七个模型（含 2k/4k 变体）、API 格式、访问目录与限制；不联网、不返回密钥。 |
| plan_ecommerce_imagesA | 免费生成可审阅的主图/详情页/场景/海报绘图计划。 scene: main 主图；detail 详情模块；scene 家居场景；poster 营销海报。 selling_points: 已核实事实，可为空列表；不自动编造商品信息。 count: 图片项数，详情页推荐 6 项；比例由用户决定。 reference_images: 原样传入计划，真正出图前才读取/下载并校验。 resolution: 可选显式 width/height 像素尺寸，原样传入每项用于模型路由。 |
| submit_image_tasksA | 批量登记1–12个后台图像任务并立即返回，不等待绘图/下载（会在后台收费）。 每项task_id在checkpointer/thread_id内唯一且稳定，同内容复用；任何冲突全批不新增。 每项可选model/api_format(auto)/endpoint(auto)；变体按显式resolution路由。 Flash-lite仅Gemini/Chat；nano-banana-pro仅DALL·E；gemini-nano仅DALL·E Generations/Gemini/Chat。 查询get_image_tasks；queued/preparing/running是正常状态，unknown不能盲目重交。 MCP进程关闭时未发送任务可在重启后继续，已可能发送任务不自动重试。 |
| get_image_tasksA | 查询持久任务与图片绝对路径，不重新提交上游。可按task_ids筛选或用next_cursor分页。 queued/preparing/running正常返回；failed/unknown/cancelled返回isError并保留全部状态。 checkpointer是本地SQLite字符串命名空间，不是路径、LangGraph对象或认证凭据。 |
| cancel_image_tasksA | 取消指定任务；queued/preparing不发图，running只停止本地等待并标unknown，不保证远端停止/退款。 |
| get_batch_resultA | 读取已登记批次及保存路径，不调用上游、不重发请求；未完成/失败返回 isError。 |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| ecommerce_brief | 收集家居日用品绘图所需的事实，规划后由用户决定是否出图。 |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| capabilities | 模型能力、边界与真实联调状态。 |
TDQS
Scored across 6 tools
Tools mostly cover distinct lifecycle stages: listing models, planning, submitting tasks, querying tasks, cancelling tasks, and reading batch results. The only mild ambiguity is between get_image_tasks and get_batch_result, since both retrieve persistent task output, but their descriptions distinguish per-task querying from batch-level result reading.
All tool names use a consistent snake_case verb_noun pattern: list_models, plan_ecommerce_images, submit_image_tasks, get_image_tasks, cancel_image_tasks, get_batch_result. The convention is predictable and readable throughout.
Six tools is well-scoped for an image-generation workflow server. Each tool earns its place by covering a specific stage: discovery, planning, submission, monitoring, cancellation, and batch retrieval.
The surface covers the core lifecycle from planning through submission, monitoring, cancellation, and result retrieval. Minor gaps remain around direct image download/upload or explicit retry/delete operations, though resubmission and cancellation provide workarounds.