About jsoncanvas
jsoncanvas is a remote MCP server for creating and managing JSON Canvas files, the infinite-canvas format used by Obsidian. It provides validation, editing, search, and exports, with an interactive viewer for supported clients.
Key features
- 01Create, read, and list canvases
- 02Validate against JSON Canvas 1.0
- 03Batch-edit nodes and edges
- 04Search canvas fields
- 05Export to markdown, SVG, Mermaid, or PNG
- 06Render canvases in an interactive viewer
Use cases
- 01Create and maintain Obsidian canvas files
- 02Check canvas data against the specification
- 03Search stored canvases for content
- 04Export canvases for documents or viewing
MCP support
Exposes canvas creation, validation, editing, search, and export tools, plus schema and example resources. It is a hosted remote MCP server requiring Microsoft Entra OAuth.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://jsoncanvas.mcpcentral.io/mcpclaude mcp add --transport http io-mcpcentral-jsoncanvas https://jsoncanvas.mcpcentral.io/mcp{"mcpServers":{"io-mcpcentral-jsoncanvas":{"type":"http","url":"https://jsoncanvas.mcpcentral.io/mcp"}}}Open in Cursor{"servers":{"io-mcpcentral-jsoncanvas":{"type":"http","url":"https://jsoncanvas.mcpcentral.io/mcp"}}}Open in VS Code{"mcpServers":{"io-mcpcentral-jsoncanvas":{"type":"http","url":"https://jsoncanvas.mcpcentral.io/mcp"}}}{"mcpServers":{"io-mcpcentral-jsoncanvas":{"serverUrl":"https://jsoncanvas.mcpcentral.io/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
https://jsoncanvas.mcpcentral.io/mcpAuth requiredTools
Details
From the project READMEJSON Canvas MCP Server
A remote Model Context Protocol server for working with
JSON Canvas files — the open infinite-canvas format used by
Obsidian. Create, validate, read, edit, export, and
search .canvas files, with an inline interactive canvas viewer.
Hosted by MCPCentral at https://jsoncanvas.mcpcentral.io/mcp
(Streamable HTTP). This repository is the public listing for the
MCP Registry; the server itself is closed-source.
mcp-name: io.mcpcentral/jsoncanvas
Authentication
The endpoint requires Microsoft Entra OAuth via the MCPCentral gateway
(login.mcpcentral.io). MCP clients perform the standard OAuth flow automatically: an
unauthenticated request returns 401 with a WWW-Authenticate pointing at the
RFC 9728 metadata at
/.well-known/oauth-protected-resource, and the client takes you through sign-in. Canvases are
stored privately per authenticated user.
Client configuration
Claude Code
claude mcp add jsoncanvas --transport http https://jsoncanvas.mcpcentral.io/mcp
Claude Desktop / other clients via mcp-remote
{
"mcpServers": {
"jsoncanvas": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://jsoncanvas.mcpcentral.io/mcp"]
}
}
}
Any MCP client that supports remote Streamable HTTP servers with OAuth (Claude, Goose, OpenAI connectors, MCP Inspector) can connect.
Tools
| Tool | Description |
|---|---|
create_canvas | Create a canvas from nodes (+ optional edges); stored under a date-prefixed .canvas name. |
validate_canvas | Validate canvas data against the JSON Canvas 1.0 spec. |
read_canvas | Read a stored canvas and return its nodes and edges. |
list_canvases | List your stored canvases. |
edit_canvas | Atomic batch add/update/remove of nodes and edges (removing a node cascades its edges). |
export_canvas | Export to markdown, svg, mermaid, or png (full-fidelity raster via Browser Rendering). |
search_canvases | Case-insensitive field-level search across your canvases. |
create_canvas, read_canvas, and edit_canvas carry MCP Apps UI metadata so supporting hosts
render the result inline in an interactive pan/zoom viewer.
Resources
canvas://schema— JSON Schema for canvas files.canvas://examples/basic— a simple two-node example canvas.ui://canvas/viewer.html— the interactive read-only viewer (MCP Apps UI).
Security
- Per-user isolation — every canvas is scoped to the authenticated user; no cross-user reads.
- Auth required — all tool calls require a valid Entra-issued bearer token.
- Hardened — strict input/size caps, bounded search, accurate tool annotations, and an SSRF-blocked PNG renderer (the headless browser cannot be coerced into outbound requests).
- Report vulnerabilities per SECURITY.md. Please do not open public issues for security reports.
License
MIT.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.