MCP server guide (for developers)

Created: 2026-10-30Updated: 2026-10-30

Connect to the PRING MCP server, and you can ask an AI agent such as Claude to "Output this month's invoices together". It creates the documents from kintone records with PRING templates and gives you links to the files. Nothing to install: you write the connection settings (URL and API key) once. Auth, permissions and limits are the same as the public API.

The MCP server is available on the Pro plan.

Requirements

A Pro plan contract.
An API key issued on the API tab under Automation & links on the settings screen. Choose scopes for what you want to do (render and files:read to create and get documents, plus records:write to add records). See Public API guide for how to issue a key.
A kintone API token registered on the API tab for each app you use.

Connection settings

The URL to connect to is https://api.pring.tech/mcp. Send the key in the Authorization: Bearer header. Do not put the key in the URL.

Claude Code

claude mcp add --transport http pring https://api.pring.tech/mcp --header "Authorization: Bearer prk_live_xxxxxxxx"

Claude Desktop

Add it to mcpServers in the settings file (claude_desktop_config.json).

{
  "mcpServers": {
    "pring": {
      "type": "http",
      "url": "https://api.pring.tech/mcp",
      "headers": { "Authorization": "Bearer prk_live_xxxxxxxx" }
    }
  }
}

Cursor

Add it to .cursor/mcp.json (project) or ~/.cursor/mcp.json in your home folder.

{
  "mcpServers": {
    "pring": {
      "type": "http",
      "url": "https://api.pring.tech/mcp",
      "headers": { "Authorization": "Bearer prk_live_xxxxxxxx" }
    }
  }
}

When you write the key in a settings file, do not put that file in a repository you share with others. Handle the key as described in Storing keys in the Public API guide.

What you can do (examples)

What you ask	What the AI agent does
“Output this month’s invoices together”	Builds a kintone query for this month’s dates, creates the documents in bulk with `bulk_render`, waits with `get_job` until it finishes, and returns the ZIP link
“Make the quote for record 101 as a PDF”	Checks the template with `list_templates`, creates one PDF with `render`, and returns the link
“Register a purchase order with this content and make it a PDF”	Checks the fields to fill with `describe_template`, then adds the record to kintone and creates the PDF with `register_and_render`

tools

tool	Arguments	Returns	Scope
`list_apps`	None	Apps that have PRING, and their templates	Any
`list_templates`	`app`	Templates of that app (ID, name, format)	Any
`describe_template`	`app`, `template`	List of fields to fill (code, name, type, required, table columns). Does not return values	Any
`render`	`app`, `record_id`, `template`, `format` (optional)	Link to the created file (once, within 15 minutes)	`render` (link needs `files:read`)
`register_and_render`	`app`, `record` (kintone REST format), `template`, `format` (optional)	Number of the added record, and link to the file	`records:write` and `render`
`bulk_render`	`app`, `template`, `record_ids` (up to 100) or `query` (kintone query), `format` (optional)	Job ID and count	`render`
`get_job`	`job_id`	Job status (`queued` / `running` / `done` / `failed`), count, and the ZIP link when done	`render` (link needs `files:read`)

query is passed to kintone as is. It can be up to 2,000 characters, and results are cut to 100 records. If there are no records, it returns “0 records”.
Files are not returned as content. Only a link is returned. As with the public API, the link opens once within 15 minutes.
With a key that does not have files:read, no file link is returned, and the result says so.
There are also resources (pring://openapi = the OpenAPI of the public API, pring://help/public-api = the key points of this help) and prompts (monthly_invoices = output this month’s invoices together, quote_to_order = create a purchase order from a quote).

For safety

Auth, IP address restriction, rate limits, plans and scopes are exactly the same as the public API. There is no way around them through MCP.
Calls made directly from a browser (calls with an Origin header) are refused.
Tool results can contain kintone values and template names. They are data, not instructions (each tool’s description says this too).
API keys and kintone tokens do not appear in responses or logs.

Public API guide

Frequently asked questions

Do I need to install anything?

No. The PRING MCP server runs on the internet (remote MCP, Streamable HTTP). Write the URL and API key in your AI agent's settings, and you can use it.

How is it different from the public API?

It does the same things. The MCP server runs on top of the public API, and auth, IP address restriction, rate limits, plans and scopes are the same as the public API. The difference is how you call it: with MCP, you do not write a program. You ask the AI agent in words.

Does my kintone data go to the AI agent?

As tool results, the AI agent gets the names of apps and templates, the numbers of records it added, and links to files. The record values themselves are on the AI agent's side when it adds a record. PRING's servers do not store record values. Files are not returned as content. Only a link that opens once within 15 minutes is returned.

Do the documents I create count toward my pages?

Yes. They count toward your monthly pages, the same as output from the screen or the public API. Documents created with a test key (prk_test_) do not count.

Can I connect with OAuth?

Currently you connect with an API key (Authorization: Bearer).

Written and maintained by: TechJapan LLC (PRING team) (Developer and operator) · Contact