Funnel FAQs

Question 1

Can I control which tools are exposed to my AI client?

Accepted Answer

Yes, Funnel offers flexible filtering options. You can use wildcard patterns to hide specific tools or entire categories from any connected MCP server, giving you granular control over the toolset presented to your AI.

Question 2

How does MCP Funnel optimize AI context and reduce token consumption?

Accepted Answer

It significantly reduces token consumption by enabling fine-grained filtering, exposing only the necessary tools to the AI. This can lead to a 40-60% reduction in context usage, and over 95% reduction when using the 'Hacky Discovery Mode'.

Question 3

What is MCP Funnel and what does it do?

Accepted Answer

MCP Funnel is a Model Context Protocol (MCP) proxy server that aggregates multiple MCP servers into a single interface. It allows AI clients like Claude, Gemini, or Codex to access tools from various sources simultaneously while optimizing context usage.

Question 4

How does Funnel prevent naming conflicts between tools from different servers?

Accepted Answer

MCP Funnel automatically prefixes tool names with their originating server's name (e.g., `github__create_issue`, `memory__store_memory`). This namespacing ensures unique identifiers and prevents conflicts when combining tools from various sources.

Question 5

What AI clients and MCP server types does Funnel support?

Accepted Answer

MCP Funnel integrates with AI clients that use the Model Context Protocol, including Claude Desktop, Claude Code CLI, Google Gemini, and Codex CLI. It also supports custom transports for stdio-based MCP servers, such as those run via Docker, NPX, or local binaries.

Funnel

Funnel

Key Features

Use Cases

Key Features

Use Cases