Your documents, inside every AI you use

Add Nobami as a custom connector on claude.ai — it then also works in Claude Desktop and the Claude mobile apps.

Free plans can add one custom connector. On Team/Enterprise an Owner adds it in organization settings first, then each member connects.

1. Open Claude → Customize → Connectors (claude.ai/customize/connectors).
2. Click “+” → “Add custom connector”. Name it Nobami and paste the server URL:

   https://api.nobami.com/mcp

   Copy
3. Click Add, then Connect. Sign in to Nobami, review the permissions and click Allow.
4. In a chat, open the tools menu and make sure Nobami is turned on. Try: “Find my 2026 budget sheet in Nobami and summarise it.”

Claude Agent SkillOpen Connectors ↗Help: custom connectors ↗

One command adds the Nobami MCP server; sign in once with /mcp. A plugin bundles the server and the Nobami skill.

1. Add the server (use --scope user to make it available in every project):

   claude mcp add --transport http nobami https://api.nobami.com/mcp

   Copy
2. Start claude, run /mcp, choose nobami → Authenticate, and allow access in the browser.
3. Or install the plugin: unzip nobami-claude-plugin.zip, then run these inside Claude Code.

   /plugin marketplace add ./nobami-claude-plugin
   /plugin install nobami@nobami

   Copy

.mcp.json (project scope)

{
  "mcpServers": {
    "nobami": {
      "type": "http",
      "url": "https://api.nobami.com/mcp"
    }
  }
}

Claude Code pluginClaude Agent SkillClaude Code MCP docs ↗

Add Nobami as a custom app/connector in developer mode (MCP + OAuth), or build a custom GPT whose Actions use the OpenAPI spec.

OpenAI rolls developer-mode connectors out by plan (Business, Enterprise and Edu first; other plans may be limited to read-only actions). Nobami registers ChatGPT automatically (dynamic client registration). Menu names can differ slightly between builds.

1. Open ChatGPT → Settings → Apps & Connectors → Advanced settings and turn on Developer mode.
2. Back in Apps & Connectors, click Create. Name: Nobami, MCP server URL below, Authentication: OAuth.

   https://api.nobami.com/mcp

   Copy
3. Confirm the notice about custom connectors, sign in to Nobami and click Allow.
4. In a new chat, open “+” → Developer mode (or More) and pick Nobami.

OpenAI help: developer mode ↗

Create a GPT whose Actions import the Nobami OpenAPI spec. Sign users in with OAuth (register an OAuth client for the GPT first) or use a Nobami API key for a private GPT.

Unlike ChatGPT apps/connectors (which register themselves automatically), GPT Actions need a manually registered OAuth client with ChatGPT’s callback URL and client_secret_post.

1. In ChatGPT open Explore GPTs → Create → Configure → Actions → Create new action → Import from URL:

   https://api.nobami.com/openapi.json

   Copy
2. OAuth (recommended for GPTs others use): choose Authentication → OAuth and copy the callback URL ChatGPT shows. Register a client for it:

   curl -X POST https://api.nobami.com/oauth/register \
     -H 'content-type: application/json' \
     -d '{"client_name":"My Nobami GPT","redirect_uris":["<ChatGPT callback URL>"],"token_endpoint_auth_method":"client_secret_post","grant_types":["authorization_code","refresh_token"],"response_types":["code"]}'

   Copy
3. Paste the returned client_id and client_secret, Authorization URL …/oauth/authorize, Token URL …/oauth/token, Scope “docs.read docs.write”, Token exchange method: Default (POST request).

   https://api.nobami.com/oauth/authorize
   https://api.nobami.com/oauth/token

   Copy
4. Private GPT instead: Authentication → API Key → Bearer, and paste a key from Nobami → Settings → AI connections → API keys.
5. Paste the instructions from gpt-action-instructions.txt, set the privacy policy URL to https://nobami.com/en/privacy/ and save.

GPT Action instructionsOpenAPI spec ↗OpenAI: GPT Actions authentication ↗

Register the Nobami MCP server in the Codex CLI / IDE extension and log in with OAuth.

1. Add the server and log in:

   codex mcp add nobami --url https://api.nobami.com/mcp
   codex mcp login nobami

   Copy
2. Or add it to ~/.codex/config.toml:

   [mcp_servers.nobami]
   url = "https://api.nobami.com/mcp"

   Copy
3. Optional: add the AGENTS.md snippet to your repository so Codex knows when to use Nobami.

AGENTS.md snippet (Codex & coding agents)MCP config examplesCodex MCP docs ↗

Add Nobami with gemini mcp add (Streamable HTTP, OAuth discovered automatically), or install the Nobami extension.

1. Add the server for your user:

   gemini mcp add --transport http --scope user nobami https://api.nobami.com/mcp

   Copy
2. Start gemini and authenticate:

   /mcp auth nobami

   Copy
3. Or install the extension: unzip nobami-gemini-extension.zip and run:

   gemini extensions install ./nobami

   Copy

~/.gemini/settings.json

{
  "mcpServers": {
    "nobami": {
      "httpUrl": "https://api.nobami.com/mcp"
    }
  }
}

Gemini CLI extensionGemini CLI MCP docs ↗

Use Gemini function calling with tool declarations generated from the Nobami OpenAPI spec and a Nobami API key on your server.

The Google Gen AI SDK also has experimental MCP support; with it you can pass an MCP client session for the Nobami server instead of writing declarations.

1. Create a Nobami API key in Nobami → Settings → AI connections → API keys. Give it only the scopes your app needs.
2. Download https://api.nobami.com/openapi.json and turn the operations you need into Gemini functionDeclarations tool (function) definitions — name, description and JSON-schema parameters map 1:1.
3. When the model returns a tool call, call the matching Nobami endpoint from your server with the header Authorization: Bearer <your key>, then send the JSON result back to the model.

   curl -H "Authorization: Bearer $NOBAMI_API_KEY" https://api.nobami.com/v1/...

   Copy
4. Keep the key on your server — never ship it in a web page or mobile app.

Gemini function calling ↗OpenAPI spec ↗

Connect your agent to the Nobami MCP server with the MCP onboarding wizard (Streamable HTTP + OAuth dynamic discovery).

Generative orchestration must be on for MCP tools. Power Platform custom connectors expect OpenAPI 2.0, so convert the spec if you go that route.

1. Open your agent → Tools → Add a tool → New tool → Model Context Protocol.
2. Server name Nobami, a short description, and the server URL:

   https://api.nobami.com/mcp

   Copy
3. Authentication: OAuth 2.0 → Dynamic discovery. Create, then create a connection and sign in to Nobami.

Microsoft Learn: add an MCP server ↗

Install with one click, or add .vscode/mcp.json. Copilot agent mode can then use the Nobami tools.

Install in VS Code

1. Click “Install in VS Code” (or run the command below). VS Code asks you to confirm.

   code --add-mcp '{"name":"nobami","type":"http","url":"https://api.nobami.com/mcp"}'

   Copy
2. Or create .vscode/mcp.json in your workspace:

   {
     "servers": {
       "nobami": {
         "type": "http",
         "url": "https://api.nobami.com/mcp"
       }
     }
   }

   Copy
3. Start the server from the mcp.json code lens, sign in to Nobami in the browser, then open Copilot Chat in Agent mode.

VS Code MCP configuration ↗

One-click install link, or add the server to ~/.cursor/mcp.json.

Add to Cursor

1. Click “Add to Cursor” and confirm the install in Cursor.
2. Or add to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project):

   {
     "mcpServers": {
       "nobami": {
         "url": "https://api.nobami.com/mcp"
       }
     }
   }

   Copy
3. Open Cursor Settings → MCP, click Connect / Login next to nobami and allow access.

Cursor install links ↗

Add a serverUrl entry to mcp_config.json and refresh Cascade’s MCP list.

On Teams/Enterprise an admin may need to allow remote MCP servers.

1. Open Cascade → MCP servers → Configure (opens ~/.codeium/windsurf/mcp_config.json) and add:

   {
     "mcpServers": {
       "nobami": {
         "serverUrl": "https://api.nobami.com/mcp"
       }
     }
   }

   Copy
2. Save, press Refresh, and sign in to Nobami when asked.

Windsurf MCP docs ↗

Based on the vendor’s changelog and partner guides; menu names may differ slightly.

Add a remote context server by URL; Zed runs the MCP OAuth flow itself.

1. Open Settings → AI → MCP Servers → Add server, or add to settings.json:

   {
     "context_servers": {
       "nobami": {
         "url": "https://api.nobami.com/mcp"
       }
     }
   }

   Copy
2. Zed opens the Nobami sign-in page; click Allow. The tools appear in the Agent panel.

Zed MCP docs ↗

Add Nobami under MCP Servers → Remote Servers with the Streamable HTTP transport.

1. In Cline open MCP Servers → Remote Servers. Name: nobami, URL below, transport: Streamable HTTP.

   https://api.nobami.com/mcp

   Copy
2. Or edit cline_mcp_settings.json:

   {
     "mcpServers": {
       "nobami": {
         "type": "streamableHttp",
         "url": "https://api.nobami.com/mcp"
       }
     }
   }

   Copy
3. Sign in to Nobami when Cline opens the browser.

Cline: remote servers ↗

Paid plans can add custom remote MCP connectors with OAuth.

Available on Pro, Max and Enterprise (admins control Enterprise). Mobile apps may only show built-in connectors.

1. Open Settings → Connectors → Add connector → Custom → Remote.
2. Enter the server URL and choose OAuth:

   https://api.nobami.com/mcp

   Copy
3. Sign in to Nobami and allow access, then enable Nobami from the sources menu in a thread.

Perplexity ↗

Based on the vendor’s changelog and partner guides; menu names may differ slightly.

Add Nobami as a Custom MCP Connector; Le Chat detects OAuth automatically.

Custom connectors are admin-only on team plans (on Free/Pro you are the admin).

1. Open Le Chat → Intelligence → Connectors → Add connector → Custom MCP Connector.
2. Name: nobami (no spaces), server URL below, then Connect.

   https://api.nobami.com/mcp

   Copy
3. Sign in to Nobami and allow access; enable the connector in a chat.

Le Chat ↗

Based on the vendor’s changelog and partner guides; menu names may differ slightly.

Give an AI Agent node the Nobami tools with the MCP Client Tool node, or call the REST API with an API key.

1. Add an MCP Client Tool to your AI Agent node. Endpoint below, transport HTTP Streamable, authentication MCP OAuth2 (n8n registers itself automatically).

   https://api.nobami.com/mcp

   Copy
2. For plain workflows use an HTTP Request node with header Authorization: Bearer <Nobami API key>; endpoints are in the OpenAPI spec.

n8n MCP credentials ↗OpenAPI spec ↗

Call the Nobami REST API from the HTTP “Make a request” module with an API key.

1. Create a Nobami API key with only the scopes the scenario needs.
2. Add HTTP → Make a request. URL: an endpoint from the OpenAPI spec; header Authorization: Bearer <key>; body type JSON.
3. Store the key in a Make connection or data store, not in plain module fields shared with others.

OpenAPI spec ↗

Use Webhooks by Zapier (Custom Request) with a Nobami API key to create or update documents from any Zap.

1. Create a Nobami API key with the scopes the Zap needs.
2. Add the action Webhooks by Zapier → Custom Request. Method and URL from the OpenAPI spec, header Authorization: Bearer <key>, data as JSON.

OpenAPI spec ↗

PlayMCP is Kakao’s open MCP platform (beta since July 2025). Its toolbox lets you use registered MCP tools in ChatGPT and Claude with a single Kakao sign-in.

Not verified: we could not confirm that Nobami is listed on PlayMCP yet. Until it appears there, connect Nobami directly with its MCP URL in ChatGPT or Claude.

1. Sign in at playmcp.kakao.com with your Kakao account and search for “Nobami”.
2. If it is listed, add it to your 도구함 (toolbox) and sign in to Nobami when asked.
3. Connect the toolbox: in ChatGPT developer mode or as a Claude custom connector use the PlayMCP gateway URL:

   https://playmcp.kakao.com/mcp

   Copy
4. Not listed yet? Use the Nobami MCP URL directly with the ChatGPT or Claude guide above.

   https://api.nobami.com/mcp

   Copy

PlayMCP ↗Kakao announcement (toolbox) ↗

HyperCLOVA X supports function calling (Chat Completions v3, April 2025) and an OpenAI-compatible API. Declare Nobami operations as functions and call the API with a Nobami key.

We found no official MCP client in CLOVA Studio as of 2026-10, so MCP connectors are not listed here.

1. Create a Nobami API key in Nobami → Settings → AI connections → API keys. Give it only the scopes your app needs.
2. Download https://api.nobami.com/openapi.json and turn the operations you need into CLOVA Studio Chat Completions v3 tool (function) definitions — name, description and JSON-schema parameters map 1:1.
3. When the model returns a tool call, call the matching Nobami endpoint from your server with the header Authorization: Bearer <your key>, then send the JSON result back to the model.

   curl -H "Authorization: Bearer $NOBAMI_API_KEY" https://api.nobami.com/v1/...

   Copy
4. Keep the key on your server — never ship it in a web page or mobile app.

CLOVA Studio function calling ↗OpenAPI spec ↗

Solar models (e.g. solar-pro4) take OpenAI-style tools at https://api.upstage.ai/v1. Generate the tools from the Nobami OpenAPI spec.

1. Create a Nobami API key in Nobami → Settings → AI connections → API keys. Give it only the scopes your app needs.
2. Download https://api.nobami.com/openapi.json and turn the operations you need into Solar tools tool (function) definitions — name, description and JSON-schema parameters map 1:1.
3. When the model returns a tool call, call the matching Nobami endpoint from your server with the header Authorization: Bearer <your key>, then send the JSON result back to the model.

   curl -H "Authorization: Bearer $NOBAMI_API_KEY" https://api.nobami.com/v1/...

   Copy
4. Keep the key on your server — never ship it in a web page or mobile app.

Upstage function calling ↗OpenAPI spec ↗

EXAONE 4.0 supports tool calling. Serve it with vLLM (OpenAI-compatible, --enable-auto-tool-choice --tool-call-parser hermes) or FriendliAI and pass Nobami tools.

1. Create a Nobami API key in Nobami → Settings → AI connections → API keys. Give it only the scopes your app needs.
2. Download https://api.nobami.com/openapi.json and turn the operations you need into EXAONE tools tool (function) definitions — name, description and JSON-schema parameters map 1:1.
3. When the model returns a tool call, call the matching Nobami endpoint from your server with the header Authorization: Bearer <your key>, then send the JSON result back to the model.

   curl -H "Authorization: Bearer $NOBAMI_API_KEY" https://api.nobami.com/v1/...

   Copy
4. Keep the key on your server — never ship it in a web page or mobile app.

EXAONE 4.0 model card ↗OpenAPI spec ↗

No public MCP connector or custom-tool option for end users could be found as of 2026-10.

1. Check Wrtn’s settings for an “MCP”, “connector” or “tool” option. If one appears, use the Nobami MCP URL with OAuth.

   https://api.nobami.com/mcp

   Copy
2. Until then, use Nobami’s built-in AI copilot or one of the supported AIs above.

Wrtn ↗

No public MCP or third-party tool connection for A. could be found as of 2026-10 (A. Biz’s agent builder is enterprise-only).

1. Enterprise A. Biz customers can ask SKT whether the agent builder accepts a remote MCP server; give them the Nobami MCP URL.

   https://api.nobami.com/mcp

   Copy
2. Individuals: use Nobami’s built-in AI copilot or one of the supported AIs above.