# Redis Context Gateway

```json metadata
{
  "schema_version": 2,
  "title": "Redis Context Gateway",
  "description": "Connect agents to MCP servers and tools over Redis data through one endpoint.",
  "categories": ["docs","develop","ai"],
  "tableOfContents": {"sections":[{"id":"what-is-context-gateway","title":"What is Context Gateway?"},{"id":"why-use-context-gateway","title":"Why use Context Gateway?"},{"id":"concepts","title":"Concepts"},{"id":"quick-example","title":"Quick example"},{"id":"manage-gateways-and-upstreams","title":"Manage Gateways and Upstreams"},{"id":"serve-tools-at-runtime","title":"Serve tools at runtime"}]}

,
  "codeExamples": []
}
```


## What is Context Gateway?

Redis Context Gateway is a Model Context Protocol (MCP) gateway for AI agents that:

- **Connects MCP servers:** Register third-party and in-house MCP servers that support Streamable HTTP as Upstreams.
- **Provides one endpoint per Gateway:** Connect your agent to a Gateway to access tools from multiple sources.
- **Manages Upstream credentials:** Configure credentials on the Gateway so your agent does not need credentials for each connected MCP server.
- **Makes Redis data available as tools:** A Context Surface generates tools from a data model describing records in Redis.
- **Scales to large tool sets:** In search mode, the agent searches for tools instead of loading every definition.
- **Handles large and repeated results:** An optional cache serves repeated calls, and compression keeps large results out of the agent's context.

<div class="grid grid-cols-1 md:grid-cols-4 gap-6 my-8">
  
  
  
  
</div>

## Why use Context Gateway?

<div class="grid grid-cols-1 md:grid-cols-2 gap-6 my-6">
  <div class="p-5 border border-redis-pen-300 rounded-lg">
    <h3 class="text-redis-ink-900 font-semibold mb-3">For AI applications</h3>
    <ul class="space-y-1 text-redis-pen-600">
      <li>Access tools from multiple sources through one MCP connection</li>
      <li>Tool definitions that fit the context window, through search mode</li>
      <li>Large results read in pages instead of filling the context</li>
      <li>Sign-in with your identity provider for MCP clients that support it</li>
    </ul>
  </div>
  <div class="p-5 border border-redis-pen-300 rounded-lg">
    <h3 class="text-redis-ink-900 font-semibold mb-3">For platform teams</h3>
    <ul class="space-y-1 text-redis-pen-600">
      <li>Upstream credentials stored once, in Vault, AWS Secrets Manager, or encrypted Redis</li>
      <li>Per-Gateway access through Identity Service agent keys</li>
      <li>Python package, <code>iris</code> command-line interface (CLI), and REST API for administration</li>
      <li>Managed deployment on Redis Cloud or self-managed deployment on Kubernetes</li>
    </ul>
  </div>
</div>

## Concepts

A **Gateway** is one MCP endpoint for an agent. Each **Upstream** behind it is an MCP server or a **Context Surface**, which generates tools from a data model of records in Redis. In full mode the agent lists every tool; in search mode it searches for the tools it needs. See [Context Gateway concepts](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/concepts) for tool policies, caching, and compression.

## Quick example

1. Set up a Gateway:
    - **Redis Cloud:** [Create a Context Gateway](https://redis.io/docs/latest/operate/iris/context-gateway/create-gateway).
    - **Self-managed:** Follow the [deployment guide](https://redis.io/docs/latest/operate/iris/context-gateway/self-managed), then [create a Gateway and an agent key](https://redis.io/docs/latest/operate/iris/context-gateway/self-managed/api-examples). A license key is required for self-managed deployment. [Contact Redis](https://redis.io/contact/) to request access.
1. Install the Redis Iris Python package, which includes the `iris` CLI:

    ```bash
    pip install 'redis-iris>=1.0.0'
    ```

1. With `IRIS_CONTEXT_GATEWAY_URL` and `IRIS_CONTEXT_GATEWAY_API_KEY` in your environment, find a tool and call it:

    ```python
    from redis_iris import ContextGateway

    gateway = ContextGateway.from_env()

    for tool in gateway.search("filter providers by specialty"):
        print(tool.name, tool.description)

    result = gateway.call("<tool-name>", {"<argument>": "<value>"})
    ```

    From the CLI, `iris --json context-gateway tools list` lists the same tools.

Use the tool names and input schemas returned by discovery. The [redis-iris package](https://pypi.org/project/redis-iris/) covers MCP calls and Gateway administration in Python and the CLI. For JavaScript and TypeScript, install the [@redis-iris/context-gateway SDK](https://www.npmjs.com/package/@redis-iris/context-gateway); see the [client guide](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/mcp#choose-a-client) to choose a client. The [quickstart](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/quickstart) shows calls in full and search mode.

## Manage Gateways and Upstreams

- [Manage Gateways and Upstreams](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/manage-gateways): create Gateways, register MCP servers, store their credentials, and set tool policies.
- [Manage Context Surfaces](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/manage-context-surfaces): generate tools for querying records in Redis.

## Serve tools at runtime

- [Connect with MCP](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/mcp) from any Streamable HTTP client.
- [Integrate with an agent](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/agent-integration) using OpenAI Agents SDK or LangGraph.

Use the [MCP API Reference](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/api-reference) for MCP tools and metadata, and [limits and errors](https://redis.io/docs/latest/develop/ai/context-engine/context-gateway/limits) for constraints and recovery guidance.