# Add ngrok skills to your coding agent

> Install eight new skills that teach Claude, Cursor, and other coding agents to secure endpoints, receive webhooks, test MCP servers, and troubleshoot ngrok.

- Date: 2026-09-19
- Shipped by: Nijiko Yonskai
- Products: ngrok Agent, Gateway
- URL: https://ngrok.com/changelog/agent-skills-launch

Your coding agent already writes the code. Now it can run ngrok too.

The [`ngrok/skills`](https://github.com/ngrok/skills) repo packages eight new skills that teach Claude, Cursor, and other coding agents to configure ngrok for common jobs. Agents can:

- Put OAuth, rate limiting, IP rules, or a custom response in front of an endpoint with **secure-endpoint**.
- Receive and verify webhooks from Stripe, GitHub, Twilio, and 70+ other providers with **receive-webhooks**.
- Expose an in-development MCP server so Claude, OpenAI, or another provider can connect to it with **run-mcp-gateway**.
- Give every sandbox, container, or tenant its own isolated endpoint with **provision-sandbox-access**.
- Diagnose common errors and connection failures with **troubleshoot-ngrok**.
- Authenticate ngrok and pick the right setup (CLI, SDK, API, or infrastructure as code) with **ngrok-setup**.

The other two, **ngrok-engine** and **ngrok-surfaces**, are reference material on endpoints, Traffic Policy, and each configuration surface. The task skills load them automatically.

`expose-localhost` now tunnels raw TCP too, so an agent can wire up SSH, RDP, and database connections, not just HTTP.

## Get started

1. Run `npx skills add ngrok/skills` in your project to install every skill, or add `--skill <name>` for just one.
2. Ask your coding agent to do something like "expose my local server" or "put OAuth in front of my endpoint."
3. Browse the full list and source on [github.com/ngrok/skills](https://github.com/ngrok/skills).
