mcpField
J
JsonFabrica

JsonFabrica

Local
jsonfabrica.comDeveloper Tools0Updated 3w ago

About JsonFabrica

JsonFabrica is a local MCP server for coding agents to generate schema-conformant synthetic JSON test data through the JsonFabrica API. It supports reproducible generation and referential integrity across related records.

Key features

  1. 01Create, manage, and reuse data templates
  2. 02Generate template-based or ad hoc JSON
  3. 03Generate batches with linked records
  4. 04Manage sequences and check usage
  5. 05Inspect batch status

Use cases

  1. 01Create customer and order test data
  2. 02Build reproducible test fixtures
  3. 03Generate batches of related records

MCP support

The server exposes tools to MCP clients over stdio and calls the configured JsonFabrica REST API. Clients can manage templates, sequences, and batches, and generate test data.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"ms-methos-jsonfabrica": {
"command": "npx",
"args": [
"-y",
"@jsonfabrica/mcp-server"
],
"env": {
"JSONFABRICA_API_KEY": "<JSONFABRICA_API_KEY>",
"JSONFABRICA_API_URL": "<JSONFABRICA_API_URL>"
}
}
}
}
Run commandnpx -y @jsonfabrica/mcp-server
Environment variables to set
JSONFABRICA_API_KEYSecretJsonFabrica API key. Create one at jsonfabrica.com -> Settings -> API Keys.
JSONFABRICA_API_URLBase URL of the JsonFabrica gateway. Override only when self-hosting the gateway.

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npm@jsonfabrica/mcp-server0.3.2

Tools

18

Saved tool list · updated 3w ago

This tool takes no parameters.

Details

From the project README

@jsonfabrica/mcp-server

npm license

A local Model Context Protocol (MCP) server that lets an AI coding agent — Claude Desktop, Cursor, or anything else that speaks MCP — generate realistic, schema-conformant JSON test data mid-session by calling JsonFabrica's REST API as MCP tools.

It runs over stdio transport only: your AI client launches it as a subprocess, so there's no network service to host and no port to open. It never talks to anything except the JsonFabrica gateway you configure.

New to JsonFabrica? It's an API-first service for generating synthetic JSON test data from reusable templates — deterministic, seed-reproducible, with referential integrity across related records. See jsonfabrica.com and the API docs.

Install / run

Add it to your MCP client's config so the client launches it via npx:

{
  "mcpServers": {
    "jsonfabrica": {
      "command": "npx",
      "args": ["-y", "@jsonfabrica/mcp-server"],
      "env": {
        "JSONFABRICA_API_KEY": "sk_live_...",
        "JSONFABRICA_API_URL": "https://api.jsonfabrica.com"
      }
    }
  }
}
  • Claude Desktop — add the block above to claude_desktop_config.json (Settings → Developer → Edit Config), then restart.
  • Cursor — Settings → MCP → Add new MCP server, or add the block to ~/.cursor/mcp.json.

Get an API key by signing up at jsonfabrica.com — Settings → API Keys.

Or run it directly for local testing:

npm install
npm run build
JSONFABRICA_API_KEY=sk_live_... node dist/index.js

What it looks like in a session

Once connected, an agent can do things like:

You: Generate 20 realistic customer records and 60 orders linked to them, and drop them into fixtures/seed.json.

Agent: calls jsonfabrica_create_template for the customer and order shapes, then jsonfabrica_create_batch with a relations map so each order references a generated customer id, then writes the result to the file.

No tab-switching to a dashboard, no hand-written fixtures.

Configuration

Env varRequiredDefaultNotes
JSONFABRICA_API_KEYYes—If missing, the server still starts and answers tool discovery, but every tool call fails with "check JSONFABRICA_API_KEY". Never logged or echoed back in tool output.
JSONFABRICA_API_URLNohttps://api.jsonfabrica.comBase URL of the JsonFabrica gateway. Override this if you're self-hosting the gateway (e.g. http://localhost:4000).

Tools

Every tool's description states, verbatim, which REST endpoint it calls. Tool names are prefixed jsonfabrica_ to avoid collisions with other MCP servers your client may have loaded.

Health / Auth
ToolEndpointNotes
jsonfabrica_healthGET /healthNo auth. Connectivity check.
jsonfabrica_whoamiGET /v1/whoamiReturns { tenantId, role } for the configured key.
Templates
ToolEndpoint
jsonfabrica_create_templatePOST /v1/templates
jsonfabrica_list_templatesGET /v1/templates
jsonfabrica_get_templateGET /v1/templates/{templateId}
jsonfabrica_update_templatePUT /v1/templates/{templateId}
jsonfabrica_delete_templateDELETE /v1/templates/{templateId}
jsonfabrica_generate_from_templatePOST /v1/templates/{templateId}/generate
jsonfabrica_generate_adhocPOST /v1/templates/generate (no persistence; same billing as persisted generation)

Template body strings use JsonFabrica's function-call placeholder syntax with angle brackets, e.g. <getRandomFullName()>, <getRandomEmail()>, <createSeq('orderNo')>. The full catalogue of built-in functions is documented at jsonfabrica.com/docs/functions — it's authoring reference, not something this server exposes as tools.

Sequences
ToolEndpoint
jsonfabrica_create_sequencePOST /v1/sequences
jsonfabrica_list_sequencesGET /v1/sequences
jsonfabrica_get_sequenceGET /v1/sequences/{name}
jsonfabrica_update_sequencePATCH /v1/sequences/{name}
jsonfabrica_delete_sequenceDELETE /v1/sequences/{name}
jsonfabrica_bump_sequencePOST /v1/sequences/{name}/bump
Batches
ToolEndpoint
jsonfabrica_create_batchPOST /v1/batches (small batches run synchronously — 200 with results; larger batches are queued — 202, poll with jsonfabrica_get_batch)
jsonfabrica_get_batchGET /v1/batches/{batchId}

Both tools return the same shape: { batchId, status, seed, results?, error? }, where status is queued | running | completed | failed. results is present once the batch is completed and groups the generated documents by alias — each value is an array of the documents themselves:

{
  "batchId": "4f1c…",
  "status": "completed",
  "seed": 42,
  "results": {
    "customer": [{ "id": "c-1", "name": "Ada Lovelace" }],
    "order": [{ "id": "o-1", "customerId": "c-1" }, { "id": "o-2", "customerId": "c-1" }]
  }
}

A failed batch carries error: { code, message } instead of results. Batches under 50 documents in total (the API's default threshold) run synchronously; larger ones return 202 { batchId, status: "queued", seed }.

Usage
ToolEndpoint
jsonfabrica_get_usageGET /v1/usage — returns { tenantId, usageTotal, asOf }
Explicitly out of scope
  • POST /v1/signup and PATCH /v1/billing/tier — unauthenticated account-creation / billing-tier-change endpoints. Wrapping these would let a model create real paid subscriptions or change billing tiers on the user's behalf; excluded by design (generation API surface only).
  • POST /v1/webhooks/stripe — Stripe-only webhook ingestion, not a developer-facing capability.

Error handling

Every tool catches errors internally and returns an MCP isError: true result with a readable message — it never throws out of the handler or crashes the host process. Common cases:

  • 401 → "Invalid or missing API key — check JSONFABRICA_API_KEY."
  • 403 → the API key's tenant is not allowed to perform the call.
  • 402 (blocked account) → the upstream blockReason is passed through.
  • Network failure (gateway unreachable) → a message naming the configured JSONFABRICA_API_URL.
  • Anything else → JsonFabrica API error [CODE] (HTTP status): message.

Troubleshooting

  • 401 / "Invalid or missing API key" → check JSONFABRICA_API_KEY is set and valid.
  • Connection refused / UPSTREAM_UNREACHABLE → check JSONFABRICA_API_URL and that the gateway is actually running and reachable from wherever this process runs.
  • 403 / FORBIDDEN → your API key's tenant is not permitted to perform that call.

Development

npm install
npm run build   # tsc -p tsconfig.json
npm test        # build + node --test dist/
npm start       # node dist/index.js (requires JSONFABRICA_API_KEY)

Source layout mirrors the OpenAPI spec's tags: one file per tag under src/tools/. src/client.ts is the only place that knows about fetch, the base URL, and the Authorization header.

License

MIT — see LICENSE.

Comments

Markdown supported

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.