tesserae-mcp
Enables AI agents to build freeform dashboards for e-ink panels by listing widgets and devices, laying out a canvas, rendering a preview, and pushing to the panel.
README
tesserae-mcp
The MCP bridge for Tesserae. It lets an AI agent (Claude Desktop, Claude Code, or any MCP client) build freeform (canvas) dashboards for your e-ink panels: it lists your widgets and devices, lays out a canvas, renders a preview to check its own work, and pushes to a panel.
This is a thin stdio client. It talks to a running Tesserae over its /api/mcp
HTTP surface, so the rendering, widgets, and devices all come from your own
Tesserae instance.
create_canvas_page → set_canvas → render_preview → (look) → set_canvas → …
Prerequisites
A running Tesserae with the MCP API enabled: Settings → System → MCP → Enable MCP API. If this bridge runs on a different machine from Tesserae, also Regenerate token there and copy it.
Install
Run this on the machine where your agent runs (your laptop/desktop), which may be different from where Tesserae runs.
pip install git+https://github.com/dmellok/tesserae-mcp
That gives you the tesserae-mcp command.
Configure your agent
Point your MCP client at tesserae-mcp. Example (Claude Desktop / Claude Code
mcpServers config):
{
"mcpServers": {
"tesserae": {
"command": "tesserae-mcp",
"env": {
"TESSERAE_URL": "http://127.0.0.1:8765",
"TESSERAE_MCP_TOKEN": "<your-token>"
}
}
}
}
TESSERAE_URL— where your Tesserae is reachable (defaulthttp://127.0.0.1:8765). For a Docker/Home Assistant install, use its LAN address, e.g.http://192.168.1.50:8765(the port must be reachable; HA ingress-only setups won't work).TESSERAE_MCP_TOKEN— the token from Settings. Omit it when the agent and Tesserae share a machine (loopback is trusted).
Then just ask: "Build me an 800×480 dashboard with the time, today's weather for Melbourne, and my next calendar event, then show me a preview."
Tools
| Tool | What it does |
|---|---|
list_widgets |
Every placeable widget (with fragments) + theme/style/font options |
get_widget_options |
A widget's options + format hints (big choice lists omitted by default) |
get_widget_choices |
Page through one option's choice rows (HA entity pickers etc.) |
probe_widget_data |
A widget's data + data_source (live/sample/error) + bindable field paths |
list_devices |
Registered panels: dimensions + colour capability (palette, mono flag) |
list_pages |
Existing canvas dashboards |
create_canvas_page |
Create an empty canvas (size it to your panel) |
get_canvas |
Read a canvas document (returns a rev for concurrency-safe writes) |
set_canvas |
Replace a canvas document (422 with field errors if invalid) |
add_element |
Append one element (live-updates an open editor) |
update_element |
Change one element in place (no full re-send) |
delete_element |
Remove one element |
patch_canvas |
Change document-level fields (size, theme, bg) without touching elements |
arrange |
Compute aligned grid/row/column boxes so you lay out by intent, not pixels |
measure_text |
Measure rendered text width/height so a box fits its content |
render_report |
Read back what rendered (values, overflow, live-vs-sample, colours) as JSON |
render_preview |
Render the canvas to a PNG the agent can see |
push_to_device |
Push the canvas to explicit device(s) |
Writes (set_canvas, add_element, update_element, delete_element, patch_canvas)
accept an optional base_rev (from get_canvas); if the page changed since, the
write returns HTTP 409 so you re-read instead of clobbering a concurrent edit.
Run without installing
From a clone:
pip install mcp
TESSERAE_URL=http://127.0.0.1:8765 python -m tesserae_mcp
License
AGPL-3.0-or-later, matching Tesserae.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。