doubao-image-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| DSH_HOME | No | Harness user directory used by setup when locating the desktop ego runtime. | |
| DOUBAO_CDP_URL | No | Explicitly specifies the local Chrome CDP address, and takes precedence over the ego state file. | |
| DOUBAO_EGO_CLI | No | Path to ego-browser.mjs used by npm run setup. | |
| DOUBAO_BROWSER_STATE | No | Custom absolute path to the ego browser.json state file. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
通过已登录豆包浏览器生图。先使用 doubao_status 检查状态。遇到 HUMAN_REQUIRED 提醒用户在浏览器完成人工验证,不自动重发。模型名称必须匹配网页;seedream5.0fast 对应 Seedream 5.0 Flash。seedream5.0Pro 对应 Seedream 5.0 Pro,需会员且 allow_pro=true;保持请求的模型版本。任意自定义比例以文本传入,返回图片实际宽高才是尺寸证据。reference_images 是本机绝对路径数组。生成结束后向用户展示返回的本地图片路径。
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| doubao_statusB | 检查豆包浏览器、登录或验证状态,返回支持的调用模型参数。会创建独立豆包标签页。 |
| doubao_generate_imageA | 通过豆包网页生成一张图片,支持参考图、Seedream模型和自动/任意文本比例。一次提交,不自动重发;返回本地图片、实际宽高、网页模型证据、对话链接和诊断。 |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 2 tools
The two tools have clearly distinct roles: one reports browser/login/model status, the other generates an image. An agent can easily decide which to call first (status) and which performs the work (generate).
Both names follow a consistent doubao_<verb_or_noun> snake_case pattern with the same server prefix. The naming is predictable and readable.
Two tools is minimal but justified for a narrowly scoped browser-driven image generation server: a prerequisite status check plus the core generation action. It is slightly thin, with no room for batch or model-listing operations.
Core capability (status check + single image generation with reference images and ratios) is covered, but there is no batch generation, no separate model/parameter discovery tool, and no job polling or cancellation for long-running generations. Agents can work around this by repeating generate calls and reading status, but the lifecycle is only partially covered.