Coding agents can speed up Nango integration work in two places: wiring Nango into your product and building Nango Functions.
This page is about configuring Nango with your coding agent. To expose Nango integrations to your product’s agents, see Agent Sessions.
Management MCP server
Give your coding agent tools to manage your Nango account and search the Nango docs.
OAuth (recommended)
OAuth authorization follows your Nango user permissions across the environments in your account. The Management MCP server uses Client ID Metadata Documents (CIMD) and doesn’t support Dynamic Client Registration (DCR).
Refer to the Management MCP reference for the full tools list and required permissions.
Claude Code
Codex
VS Code
Add the server for your user so it is available across projects:Complete the Nango authorization in the browser, then run claude mcp list to verify the connection. Add the following to ~/.codex/config.toml for a global configuration, or .codex/config.toml for a project configuration:Start the OAuth flow:Complete the Nango authorization in the browser, then use /mcp to verify the connection. Run MCP: Open User Configuration from the Command Palette, then add the server:Select Start above the server configuration. VS Code opens the Nango authorization page in your browser. Complete the authorization, then check that the server is running.
After connecting, ask the agent to call environments_list, then use one of the returned environment names for subsequent tool calls.
API key
An Environment API key is useful for trusted clients when OAuth is not suitable, for example when the client does not support CIMD yet. Each key selects one environment, so these clients do not use environments_list or pass an environment argument. Create a key in Environment Settings > API Keys, replace <NANGO-API-KEY> in the relevant configuration below, and limit its scopes to the tools the client needs.
Claude Code
Cursor
Codex
VS Code
Pi
OpenCode
Add the server for your user so it is available across projects:Run claude mcp list to verify the connection. Add the following to ~/.cursor/mcp.json for a global configuration, or .cursor/mcp.json for a project configuration:Restart Cursor after saving the file. Add the following to ~/.codex/config.toml for a global configuration, or .codex/config.toml for a project configuration:Restart Codex after saving the file, then use /mcp to verify the connection. Run MCP: Open User Configuration from the Command Palette, then add the server and a secure input for the API key:VS Code prompts for the API key when it first starts the server and stores the value securely. Pi does not include an MCP client, so first install the pi-mcp-adapter extension:Then add the following to ~/.config/mcp/mcp.json for a global configuration, or .mcp.json for a project configuration:Restart Pi after installing the extension, then use /mcp to verify the connection. Add the following to ~/.config/opencode/opencode.json for a global configuration, or opencode.json in a project root for a project configuration:Run opencode mcp list to verify the connection.
Do not commit API keys to source control. Prefer user-level configuration when storing an API key inline.
Skills
Nango’s public skills live in NangoHQ/skills.
Install the core skill to implement Nango integration functions:
See the functions guide for more details.
Docs MCP server
If your coding agent only needs documentation, connect it directly to Mintlify’s public Nango documentation MCP server:
This is an unauthenticated alternative to the documentation tools included in the Management MCP server. If you use the Management MCP, you can skip this one.
Just-in-time integrations
Nango integrations are defined as code. Usually, you or your coding agent write functions in a Git repo, review them like application code, and deploy them to Nango.
Nango also exposes a Functions API for creating integrations just in time, without asking the user to interact with a codebase. A user can describe the integration behavior in text, and an agent can turn that into function code, compile it, dry run it against a connection, and deploy it directly to Nango.
This is useful for faster onboarding, prototypes, and dynamic product experiences where integration behavior depends on what the user needs at that moment. It blurs the line between creating and consuming integrations: at consumption time, an agent can define the behavior, validate it, and make it available through Nango. See build with the Functions API for the workflow and API references.
These resources help agents discover Nango docs, provider slugs, API shapes, and runtime tool-calling paths.