# Welcome to the Chain.Love's MCP Module

Chain.Love MCP Gateway is a hosted [Model Context Protocol](https://modelcontextprotocol.io/) gateway for AI agents that need reliable Web3 context and execution tools. Instead of hardcoding RPC endpoints, manually browsing service directories, or connecting every third-party MCP server one by one, an agent can use one Chain.Love MCP endpoint to discover networks, search the registry, and proxy downstream MCP tools.

{% hint style="success" %}
The detailed landing page is available here: <https://chain.love/mcp-gateway>
{% endhint %}

## What problems does it solve?

* **Web3 data aggregation in one place**: instead of stitching together chain lists, RPC providers, explorers, oracles, ramps, and service docs from different websites, an agent gets one gateway backed by Chain.Love's Web3 dataset. The current registry snapshot covers `55` supported chains, and that dataset keeps growing every day.
* **Access to many MCP servers through one public endpoint**: clients such as VS Code, Antigravity, and other AI tools often impose practical limits on how many MCP servers you can keep connected at once. By connecting Chain.Love MCP once, your agent gets access to a large catalog of registry-listed MCP servers through a single gateway connection. You can browse the open-source list here: [mcpservers.csv](https://github.com/Chain-Love/chain-love/blob/main/references/offers/mcpservers.csv).
  * **Public hosted MCP servers**: because the hosted gateway is currently public, Chain.Love can proxy your agent directly to public hosted MCP servers without additional approval flows or setup.
  * **Private hosted MCP servers**: for private hosted MCP servers or credentialed downstream integrations such as GitHub, add the required credential header to your agent configuration and Chain.Love will be able to proxy requests to that MCP through the same gateway.
  * **Self-hosted MCP servers**: if the tool is self-hosted, Chain.Love can help the agent understand how it should be deployed in the local workspace so the agent can call it directly.
* **Less hallucination, more current context**: agents call `discover_networks`, `discover_categories`, and `search` against registry-backed data instead of guessing which chains, categories, or providers exist.

## Core capabilities

Below is the high-level product surface before you move into setup and runtime details.

{% stepper %}
{% step %}

#### Discover chains

Use `discover_networks` to get the current canonical chain keys supported by Chain.Love. The current snapshot includes `55` networks, including `ethereum`, `base`, `arbitrum`, `optimism`, `filecoin`, `solana`, `bitcoin`, `sui`, and more, and the registry keeps expanding over time.
{% endstep %}

{% step %}

#### Search the registry

Use `discover_categories` and `search` to browse categories such as `apis`, `explorers`, `ramps`, `oracles`, `bridges`, `wallets`, `security`, and `mcpservers`. This is the fastest way to find RPCs, block explorers, hosted tools, and other Web3 services without hardcoding provider names.
{% endstep %}

{% step %}

#### Proxy downstream MCP tools

Use `execute` to connect to downstream MCP servers, inspect their tools, and call them through the same Chain.Love gateway session. For example, an agent can connect a GitHub MCP, list the available tools, and run the exact action it needs without adding a separate MCP endpoint to the client.
{% endstep %}
{% endstepper %}

***

## Read next

Use the pages below as the practical next steps after the overview.

{% stepper %}
{% step %}

#### [Connect an AI agent](https://chain-love.gitbook.io/mcp-module/getting-started/connect-ai-agent)

Set up the hosted endpoint, add optional tokens or downstream credentials when needed, and test the connection.
{% endstep %}

{% step %}

#### [Tools and workflow](https://chain-love.gitbook.io/mcp-module/reference/tools-and-workflow)

Understand what `discover_networks`, `discover_categories`, `search`, and `execute` are used for.
{% endstep %}
{% endstepper %}