Semiotic MCP Server

Name: Semiotic MCP Server
Author: nteract

Pick, validate, repair and render Semiotic React charts, and audit them for accessibility

Official · nteract TypeScript Remote (hosted) · Streamable HTTP Local · stdio, Streamable HTTP Developer tools Data and analytics

GitHub npm Website

84 popularity score ★ 2.7k ↓ 9.6k/wk

Connect your client

Remote (hosted)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server.

Or add it by command or config

Its docs show no set-up for a particular client: add this address in your client the way it adds a remote MCP server.

Addressits hosted endpoint

https://semiotic-mcp-server-481507046413.us-west1.run.app/mcp

Local (once it is running)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server.

Or add it by command or config

Claude CodeTerminal

claude mcp add --transport http semiotic http://127.0.0.1:3001/mcp

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "semiotic": {
      "url": "http://127.0.0.1:3001/mcp"
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "semiotic": {
      "type": "http",
      "url": "http://127.0.0.1:3001/mcp"
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.semiotic]
url = "http://127.0.0.1:3001/mcp"

Gemini CLITerminal

gemini mcp add --scope user --transport http semiotic http://127.0.0.1:3001/mcp

Local (npm)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server.

Or add it by command or config

Claude CodeTerminal

claude mcp add semiotic -- npx -y -p semiotic semiotic-mcp

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "semiotic": {
      "command": "npx",
      "args": [
        "-y",
        "-p",
        "semiotic",
        "semiotic-mcp"
      ]
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "semiotic": {
      "command": "npx",
      "args": [
        "-y",
        "-p",
        "semiotic",
        "semiotic-mcp"
      ]
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "semiotic": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "-p",
        "semiotic",
        "semiotic-mcp"
      ]
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.semiotic]
command = "npx"
args = ["-y", "-p", "semiotic", "semiotic-mcp"]

Gemini CLITerminal

gemini mcp add --scope user semiotic npx -- -y -p semiotic semiotic-mcp

Settings 11

Set together: http
`MCP_TOOL_PROFILE`	Optional: Tool set: developer (24 tools) or public (5 workflow tools); --profile overrides it, and any other value stops startup. Default `developer`.
`MCP_MAX_TOOL_OUTPUT_BYTES`	Optional: Largest tool response sent before an error is returned instead. Default `2097152`.
`MCP_MAX_RENDER_WORK_MS`	Optional: Time budget for render work. Default `2500`. For renderChart.
`MCP_MAX_PNG_CONVERSION_MS`	Optional: Time budget for PNG conversion. Default `4000`. For renderChart.
`MCP_HOST`	Optional: HTTP mode: address to bind; --host overrides it. Default `127.0.0.1`.
`MCP_AUTH_TOKEN` secret	Optional: HTTP mode: token clients must send in the Authorization header.
`MCP_AUTH_SCHEME`	Optional: HTTP mode: scheme expected before MCP_AUTH_TOKEN. Default `Bearer`.
`MCP_ALLOWED_HOSTS`	Optional: HTTP mode: comma-separated Host header allowlist.
`MCP_ALLOWED_ORIGINS`	Optional: HTTP mode: comma-separated Origin allowlist, needed by browser clients.
`MCP_MAX_BODY_BYTES`	Optional: HTTP mode: largest request body accepted. Default `4194304`.
`OPENAI_APPS_CHALLENGE_TOKEN` secret	Optional: HTTP mode: token served at /.well-known/openai-apps-challenge for ChatGPT Apps domain verification.

What it does

Semiotic is a React data visualization library from the nteract project. Its npm package carries an MCP server, semiotic-mcp, so a coding assistant can look up, check and render Semiotic charts through tool calls instead of writing chart code unchecked.

In its default developer profile the server registers 24 tools. They cover prop schemas for any component, chart and dashboard recommendations from sample rows or stream schemas, anti-pattern diagnosis and repair of a chart configuration, and static rendering to SVG or PNG. There are also data summaries and grounding payloads so the assistant can answer questions about a chart, Chartability accessibility and mobile audits, and checks for misleading design. A further set audits an explicit artifact contract under a named policy and explains refusals. It also serves 13 resources (schemas, examples, behaviour contracts, a system prompt, task packets and a ChatGPT Apps widget template) and two prompts for building and debugging a chart.

Started with --profile public (or MCP_TOOL_PROFILE=public), it registers a smaller set of five workflow tools instead: createChart, improveChart, explainChart, auditChart and getChartSchema. Startup is strict. An unknown flag, a repeated flag or a profile name other than developer or public stops the server before it serves anything.

It runs locally with npx over stdio, or with --http for Streamable HTTP on port 3001 by default. The registry entry also lists a hosted Streamable HTTP endpoint on Google Cloud Run. No API key or account is needed.

What it needs

To run it yourself	Node.js 22.12+
Runs on	macOS, Windows, Linux

Limits

Not covered renderChart is a static snapshot: props must carry data, and ref/push-mode streaming charts cannot be rendered
Limit PNG output needs the sharp package, an optional dependency; if it did not install, only SVG works
Size cap HTTP mode refuses request bodies over 4 MB by default (MCP_MAX_BODY_BYTES)
Size cap Tool responses over 2 MiB are replaced by an error by default (MCP_MAX_TOOL_OUTPUT_BYTES)
Rate limit Render work times out after 2500 ms and PNG conversion after 4000 ms by default (MCP_MAX_RENDER_WORK_MS, MCP_MAX_PNG_CONVERSION_MS)
Not covered suggestChart takes 1 to 5 sample rows
Early release renderInteractiveChart, the ChatGPT Apps widget surface, is called experimental in the README
Limit HTTP mode rejects any browser request carrying an Origin header unless MCP_ALLOWED_ORIGINS lists it; wildcard origins are not supported
Limit An unknown or repeated startup flag, or an invalid profile name, stops the server at startup

Who it suits

Suits developers building charts with Semiotic in React who want their coding assistant to check props against the real schema, catch configuration mistakes and see a rendered result before handing over code. Everything runs on the user's machine with no account, and every tool only reads its input and returns a result. It is of little use for projects on other chart libraries, since all schemas and recommendations are Semiotic's. Teams wanting a remote connector for ChatGPT can run HTTP mode behind a tunnel with a host allowlist and bearer token, or use the hosted endpoint if sending chart data to the maintainers' service is acceptable.

Tools 29

29 only read. A ? marks an optional parameter.

getSchema: Returns a component's prop schema, or lists all components when none is namedcomponent?: string
suggestChart: Recommends a chart from one to five sample rows, with capability filtersdata: arrayintent?: stringcapabilities?: object
suggestCharts: Recommends static charts by capability, taking the audience into account
proposeChartVariants: Ranks variants of, and alternatives to, a chosen chart
suggestStreamCharts: Recommends streaming charts from a data schema
suggestStreamDashboard: Lays out a multi-panel dashboard from stream schemas
suggestDashboard: Recommends a multi-panel dashboard
suggestStretchCharts: Suggests less familiar charts suited to an audience's chart literacy
repairChartConfig: Checks a chart choice against the data and proposes alternatives
renderChart: Renders a static-data chart to SVG, or to PNG when sharp is installedcomponent: stringprops?: objectcontract?: objecttheme?: objectformat?: enum
interrogateChart: Summarises a chart's data so questions about it can be answered
groundChart: Returns a chart's description, intent and navigable structure for a non-visual reader
diagnoseConfig: Finds anti-patterns in a chart configuration
auditAccessibility: Runs a Chartability accessibility audit
evaluateChart: Evaluates data fit, misleading design and accessibility together
auditMobileVisualization: Audits a chart for mobile display
reportIssue: Builds a pre-filled GitHub issue link for a bug or feature request; it does not file it
applyTheme: Explains how to use a theme preset
renderInteractiveChart: Wraps a rendered chart SVG in a ChatGPT Apps widget
suggestTokenEncoding: Recommends semantic token and ISOTYPE encodings
auditArtifact: Audits an explicit interpretation contract against a named policy
recommendRepresentation: Weighs chart and non-chart ways to present the data
repairArtifact: Proposes repairs to an artifact or fills its missing identity fields
explainRefusal: Explains why a policy refused an artifact contract
createChart: Public profile only: picks, validates, renders and proves a chart from data rows
improveChart: Public profile only: diagnoses a configuration and proposes repairs or variants
explainChart: Public profile only: describes a chart's intent and structure without pixels
auditChart: Public profile only: runs design, accessibility and mobile checks with ranked findings
getChartSchema: Public profile only: returns prop-schema guidance for a component

Resources 13

semiotic://build-info: Package version and build details
semiotic://tasks: Index of task packets
semiotic://tasks/{taskId}: One task packet
semiotic://schema: Full component schema
semiotic://artifact-contract-schema: Schema of the artifact contract
semiotic://schema-index: Index of component schemas
semiotic://schema/{component}: One component's schema
semiotic://components: Component index with renderable status
semiotic://surface-manifest: Manifest of the AI-facing surface
semiotic://behavior-contracts: Semantic rules for colours, prop combinations and streaming
semiotic://system-prompt: Packaged guidance for assistants
semiotic://examples: Working chart examples
ui://semiotic/chart-widget.html: ChatGPT Apps widget template used by renderInteractiveChart

Prompts 2

build-semiotic-chart: Step-by-step workflow for building a chart with the tools
debug-semiotic-chart: Workflow for debugging a chart from its component, problem and props

Popularity

84 popularity score, on our list (the bar is 36)

★ 2.7k ↓ 9.6k/wk

GitHub stars	2,717	41.2 points
Stars in 30 days	+3	3 points
npm downloads, last week	9,557	31.8 points
Official server	Yes	8 points

Measured 2026-11-07. How the score works.

Where it is published

Code	nteract/semiotic ★ 2.7k
npm	semiotic 3.13.3, 2026-10-29
Hosted	`https://semiotic-mcp-server-481507046413.us-west1.run.app/mcp` Streamable HTTP · no sign-in

What it can reach

Runs	On your machine, or on the maker's servers
Reaches	Nothing beyond its own data
Changes things	No: it reads only
Sign-in	None
Licence	Apache-2.0 (npm)
Last code change	2026-11-07

Every tool is registered read-only and non-destructive in the server source, which reads only its own packaged schema and example files and makes no outbound fetch calls. reportIssue just builds a GitHub issue link. Run with npx it executes on your machine over stdio with no sign-in. HTTP mode is stateless and binds to 127.0.0.1 unless --host or MCP_HOST is given. It accepts only loopback Host values until MCP_ALLOWED_HOSTS is set, rejects any browser Origin not in MCP_ALLOWED_ORIGINS (wildcards and the null origin are never accepted), caps request bodies at 4 MB by default, and requires a token only when MCP_AUTH_TOKEN is set. The hosted endpoint runs on the maintainers' Google Cloud Run service, so chart data sent there leaves your machine.

Over time

GitHub stars 2.7k +6

2.7k2.7k

8 Oct 20267 Nov 2026

npm downloads a week 9.6k +536 (+6 %)

19k1.8k

20 Oct 20257 Nov 2026

Before the site measured it, read from npm's daily download counts.

Popularity score 84 +9 (+12 %)

8675

8 Oct 20267 Nov 2026

What is rising across the directory

Sources

Page checked 2026-11-06.

Badge for your README

Shows this server's rank in its category, and stars for a server that runs on your machine. It updates every hour.

MarkdownREADME.md

[![Semiotic MCP Server on EveryMCPServer](https://everymcpserver.com/badge/semiotic.svg)](https://everymcpserver.com/servers/semiotic)

Connect your client

Add in one click

Or add it by command or config

Add in one click

Or add it by command or config

Add in one click

Or add it by command or config

Settings 11

What it does

What it needs

Limits

Who it suits

Tools 29

Resources 13

Prompts 2

Similar servers

MCP Server Chart

nt8-mcp

Knock MCP Server

Convex MCP Server

Hygraph MCP Server

BC Code Intelligence MCP Server