ngrok gives developers and platform teams one service for getting traffic to their apps, APIs and AI models, and for securing that traffic, wherever the workloads run. The same service covers a developer testing on a laptop and a production estate spread across clouds and clusters.
The problem it addresses is that every environment usually needs its own way in, each with its own security controls. With ngrok, access rules are applied once at the edge, and the same model now extends to AI traffic and MCP servers.
At A Glance
| Vendor | ngrok |
|---|---|
| Category | Network gateway, ingress and AI gateway |
| Identities covered | Human users, automated processes, API keys, AI provider keys, tokens |
| Deployment | Global cloud service with agent software, agent CLI, SDKs and Kubernetes Operator |
| Agentic AI | AI Gateway, MCP gateway, ngrok MCP server |
| Certifications | SOC 2 Type 2, HIPAA with BAA, GDPR, CCPA, EU-US Data Privacy Framework |
| Customers | Databricks, GitHub, Calendly, Cyera |
| Headquarters | San Francisco |
| Market classification | Network Security |
What Ngrok Does
ngrok manages connectivity for laptops, clouds, Kubernetes clusters and private networks in one layer. Traffic arrives at an ngrok endpoint, passes through the customer’s security and routing rules, and is delivered to the upstream service, whether that is a container in the cloud or a process on a developer’s machine.
The result is one set of controls and one entry point for every environment, instead of a different gateway, certificate set and firewall rule base for each place a service happens to run.
Products
Gateway
Gateway provides the core building blocks: endpoints (the URLs traffic arrives on, created by an agent or in the cloud), custom domains, TLS certificate management, load balancing, WAF, DDoS protection and tunnels. It also manages credentials for the human users and automated processes that operate the account.
Share Localhost
Share Localhost is the original ngrok experience. The ngrok agent opens encrypted tunnels that deliver internet traffic to a service on the local machine, so a developer can test webhooks or preview an app without deploying it.
Traffic Policy
Traffic Policy lets teams filter, match, manage and orchestrate traffic on their endpoints. Its actions include WAF checks, rate limits, access rules, redirects and forwarding, and JWT validation that rejects bad tokens before they reach the service.
AI Gateway
AI Gateway, offered at ngrok.ai, puts one URL in front of hosted and self-hosted models. It has built-in support for OpenAI, Anthropic, Google, DeepSeek and xAI, routes to local models such as Ollama or vLLM, and fails over across models and keys.
Ngrok MCP Server
The ngrok MCP server lets an AI agent manage a customer’s ngrok account, including endpoints, domains and Traffic Policy, with sign-in through OAuth.
How It Works
ngrok runs a global cloud service that works with agent software near the upstream service. The control plane is separate from the data plane: control requests are handled from a US-based AWS region, while regional data planes route traffic through global Points of Presence in the US, Europe, Asia Pacific, Japan, India, Australia and South America. A domain can be pinned to a region for data residency.
Agent endpoints are created with the agent CLI or agent SDKs, and agent-to-edge connections use TLS 1.2 or later. A Kubernetes Operator runs ngrok inside clusters.
By default ngrok does not store traffic content. Connection metadata goes to the Event Store, and Traffic Inspector Full Capture, which keeps headers and bodies for up to 3 days, is opt-in. With end-to-end encryption, ngrok never terminates TLS.
Identities And Systems Covered
| Human users | Account access with SSO and SCIM, RBAC and audit logs |
|---|---|
| Automated processes | Credentials for scripts and services that use the platform |
| API keys and credential tokens | Stored only as one-way salted hashes |
| AI provider keys | AI Gateway access key configurations scope providers, models and credentials |
| Tokens at the edge | JWT and bearer token checks on inbound requests |
| Systems | Kubernetes clusters, local machines, and hosted and local AI models |
Agentic AI Capabilities
| Model access | Requests through AI Gateway can be paid with credits, the customer’s own provider keys or a self-hosted model, and an AI Gateway API manages keys, providers, settings and usage from code. |
|---|---|
| Exposing MCP servers | AI apps such as ChatGPT, Claude.ai and Grok call MCP servers from their own cloud, so a local MCP server needs a public, protected address. ngrok’s MCP gateway guidance applies bearer token rules, secret paths, IP category allowlists and rate limiting in ngrok’s cloud, so requests that fail the rules never reach the machine. OAuth stays in the MCP server itself, with ngrok forwarding the discovery and token requests. |
Company Background
| Founder | Alan Shreve, Founder and Chief Executive Officer |
|---|---|
| Headquarters | San Francisco, United States |
| Leadership | Heather McLinden (Chief People Officer), Peter Shafton (Chief Technology Officer), Sam Richard (Chief Revenue Officer) |
| Funding | A $50 million Series A, announced in December 2022 as the company’s first fundraise, from investors including Lightspeed Venture Partners and Coatue |
| Developers | A reported 15.3 million developers have signed up |
Note: This overview is compiled from ngrok’s publicly available information as of 3 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.