Semiotic MCP
Semiotic
Semiotic's MCP server: an agent picks the chart type, renders SVG and validates the configuration against component schemas.
- 2,717 ★ on the repository
- Last commit
- Apache-2.0
Minimal access
Why this level
- Local run with no keys; the server draws charts from supplied data.
Install
- Works in
Manual install
npx -y -p semiotic semiotic-mcpThe first MCP launch command in the README, stdio transport. The library for the project is installed separately with `npm install semiotic`.
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- Local run with no keys; the server draws charts from supplied data.
This is third-party code. Review the repository files before installing.
What it does
Semiotic, a React data visualization library, includes the `semiotic-mcp` server. The `renderChart` tool draws a chart to static SVG and returns render evidence such as mark counts so an agent can verify the result. `getSchema` returns a component's prop schema, while `suggestChart` and `suggestCharts` recommend a chart type from sample rows. `diagnoseConfig` finds empty data, bad dimensions and missing accessors, and `applyTheme` returns theme presets. The server starts over stdio with no keys and over HTTP on `127.0.0.1`.
Who it is for. React developers who have an agent build a chart or dashboard and want the result verified rather than handed over blind.
Good fit when
- An agent picks a chart type for the shape of the data.
- After a config edit someone must confirm that marks were actually drawn.
- A dashboard is assembled from streaming data with several panels.
- A theme preset such as `tufte` is brought into a project.
Not a fit when
- The project is committed to another charting library.
- Only simple bars are needed and one line of code is enough.
- HTTP mode would be exposed publicly without an allowlist.
Example request
Pick a chart for the `month` and `revenue` columns, render it with `renderChart` and report how many marks were drawn.MCP
- Transport
- stdio, http
- Authentication
- not required
Limitations
The server is tied to Semiotic components only. It draws static SVG, and some components work only in the browser and are unavailable through `renderChart`. The README's model evaluations are specific to models and tasks and do not promise first-try correctness. In HTTP mode, browser requests need an explicit `MCP_ALLOWED_ORIGINS` list. The `semiotic-ai --doctor` check catches common configuration mistakes but does not replace looking at the chart.
How to disable. Delete the `semiotic` entry from the client's MCP settings and stop the HTTP process if `--http` was used. The `semiotic` package in the project can stay.
FAQ
How do I connect the server?
Run `npx -y -p semiotic semiotic-mcp` and put the same command into the client's `mcpServers`. No keys or login are needed.
How do I start HTTP mode?
With `npx -y -p semiotic semiotic-mcp --http --port 3001`. It binds to `127.0.0.1` by default.
How do I check a configuration without MCP?
Run `npx -p semiotic semiotic-ai --doctor`: it validates component and props JSON and suggests typo fixes.
Can HTTP mode be protected?
Yes. `MCP_AUTH_TOKEN` enables a bearer token, while `MCP_ALLOWED_HOSTS` and `MCP_ALLOWED_ORIGINS` set permitted addresses.
Is the `semiotic` package needed in the project?
`npx` is enough for the server. The library itself is installed in the project separately with `npm install semiotic`.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server from AntV for generating charts and diagrams: 25+ visualization types, from lines and bars to maps, mind maps and network graphs
A visualization language for agents and an MCP server: neat charts from a simple semantic spec
MCP Dashboards: interactive charts in chat
MCP Dashboards
32 tools draw interactive charts, KPI widgets and dashboards right inside the chat in Claude Desktop, VS Code and other clients
A knowledge graph with an MCP server: the agent searches concepts by meaning and sees how well each is supported.
BI as code: YAML dashboards, a semantic layer and a local MCP server through which an agent drafts charts.
MCP server and CLI that turns paper text or data into academic figures, diagrams and plots
README in short
The README lists the `semiotic-mcp` server, the `semiotic-ai --doctor` check, the `ai/schema.json` schema and an `AGENTS.md` for repository agents.