Multi-user MCP management and lazy proxy gateway that makes MCP work like Skills: progressive capability discovery, on-demand server startup, and one stable gateway for every agent.
Wellknown found it in public sources; nobody has proven control of it yet. Claiming takes one click if the repository is under your GitHub account, or a small file on your domain otherwise. Verified owners get the badge, 15-minute checks, status alerts, edits that outrank crawled data, and a ranking boost.
Agents can do it too: POST https://wellknown.network/api/v1/claims with {"agent":"mcp-manager-gateway","method":"well_known_file"} — machine-readable steps at claim.json, guide at /docs/claim.
Everything here was measured by our prober or read from a registry. Nothing is self-reported.
Attributed to the source that supplied each field. Treated as claims, not facts.
# MCP Manager Gateway > Make MCP work like Skills: discover capabilities progressively, start servers only when needed, and give every agent one stable gateway. [](https://pypi.org/project/mcp-manager-gateway/) [](https://pypi.org/project/mcp-manager-gateway/) [](https://github.com/eraycc/mcp-manager-gateway/blob/master/LICENSE) [](https://github.com/eraycc/mcp-manager-gateway/releases) [](https://hub.docker.com/r/eraycc/mcp-manager-gateway) [](https://ghcr.io/eraycc/mcp-manager-gateway) English | [简体中文](https://github.com/eraycc/mcp-manager-gateway/blob/master/README.zh-CN.md) | [Documentation Wiki](https://github.com/eraycc/mcp-manager-gateway/blob/master/wiki/README.md) MCP servers are useful, but a large MCP configuration creates a new set of problems: - every agent connects to many servers at startup, making startup slower; - complete tool schemas consume context before the agent has done any work; - rarely used servers stay alive and consume processes and connections; - each client needs another copy of the same configuration; - agents can understand existing configuration, but users still have to recreate services one by one. MCP Manager Gateway, or **MMG**, puts those servers behind one gateway. An agent sees a small discovery surface, searches for the capability it needs, reads the exact schema, and calls the selected tool. Catalog operations do not start downstream services; a `lazy` server starts on its first real call. ~~~text Agent / IDE…
Mapped onto the structured taxonomy from declared text and observed tool names. Confidence shown for derived entries.
Every source is kept verbatim. Field changes are logged as events.