Mimir — declarative MCP router with LLM-guided onboarding, scoped plugins and a layered security model.
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":"mimir-router-mcp","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.
# Mimir [](https://github.com/CTRQuko/mimir-mcp/actions/workflows/ci.yml) [](https://pypi.org/project/mimir-router-mcp/) [](https://pypi.org/project/mimir-router-mcp/) [](LICENSE) 🇬🇧 You're reading the English version · 🇪🇸 [Léeme en español](README.es.md) > *"Mimir guarda el pozo de la sabiduría: aconseja a Odín cuando le falta > contexto. Tu router MCP hace lo mismo con el LLM."* **Mimir is a declarative MCP router** built on FastMCP 3.x. Drop a `plugin.toml` next to any MCP server and Mimir discovers it, mounts it under its own namespace, scopes its credentials, filters its tools, and exposes the union to your client (Claude Desktop, an agent in a loop, anything that speaks MCP over stdio). What sets Mimir apart from other MCP aggregators: - **Declarative plugin contract** — every plugin ships its own `plugin.toml` describing identity, runtime, security, requirements. No central config to keep in sync. - **Inventory separated from plugins** — operators describe their infrastructure in `inventory/*.yaml` (hosts, services). Plugins ask the router *"give me hosts of type X"* — they never hardcode IPs or hostnames. - **LLM-guided onboarding** — when a plugin needs hosts or credentials the operator hasn't supplied, the router exposes a `setup_<plugin>()` meta-tool. The LLM walks the operator through what's missing, conversation by conversation. - **Layered security** — manifest quarantine, audit log, scoped credential vault, profile gate, tool whitelist/blacklist, and cross-plugin env scoping in subprocess. See [`docs/security-model.md`](docs/security-model.md). ## Quick install ```bash git clone https://githu…
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.