mcp.run alternatives
A fair, factual comparison — and where to go if you want 470+ curated MCP servers with a copy-paste config for Claude, Cursor and Windsurf, a strict “official” flag, and no account.
What mcp.run is
mcp.run is a developer platform for publishing, hosting and running MCP servers — it calls them “servlets” — with a registry, secrets management and a hosted runtime you connect clients to. It's more a run/host platform than a plain browse-and-copy directory.
Pricing: Free tier for individuals; paid tiers for teams/usage. An account is required to run servlets. Visit mcp.run
Choose mcptrove if…
- You want to browse and paste a config, not publish/host via an SDK
- You want to run servers locally with no account
- You want curated discovery — capability pages, comparisons, quality signals
Choose mcp.run if…
- You're a developer who wants to host and run servlets in the cloud
- You want secrets management and a hosted runtime
- You're building on its SDK/CLI
mcp.run strengths
- Hosted runtime — no local infrastructure
- Secrets management for credentials
- Registry of servlets with a programmatic SDK/CLI
- Portable servlets across clients
Things to check
- Developer/CLI-first — oriented around publishing & hosting, not browse-and-paste
- Account required, and servlets run through its hosted runtime
- It's a run platform, not a curated discovery catalogue with quality signals
The 3 things people complain about with mcp.run
Honest answers to the most-searched concerns — what's actually true, and how mcptrove is different.
“It's developer/CLI-first, not browse-and-paste”
mcp.run is built around publishing and running servlets via its SDK/CLI (you'll see `mcp.run(transport=…)` in its docs). Powerful for developers, heavier if you just want a tool running in your client.
On mcptrove: mcptrove is browse-and-copy: find a server, paste its config for your client, done — no SDK, no CLI.
“You need an account and run through its hosted runtime”
Servlets execute on mcp.run's platform under your account — that's the hosting convenience, and also a dependency on the platform.
On mcptrove: Run servers yourself locally, with no account and no hosted runtime in the path.
“It's a run platform, not a discovery catalogue”
mcp.run's strength is hosting and running servlets; it's not built as a curated discovery surface with head-to-head comparisons and quality signals.
On mcptrove: We curate servers with capability pages, comparisons and per-client config so discovery is the whole point.
Popular servers on mcptrove
Each one ships a copy-paste config for Claude, Cursor and Windsurf, the real tool list, and quality signals.
Official MCP reference server that fetches a URL and returns its content as clean Markdown, with chunking.
Official MCP reference server for secure local filesystem read/write within allowed directories.
Official MCP server providing persistent, file-backed knowledge-graph memory across sessions.
Official MCP server for reading, searching, and manipulating a local Git repository's files and history.
Structured step-by-step reasoning tool for breaking problems into revisable thought sequences.
GitHub's official server for repos, issues, PRs, and Actions — local Docker or hosted remote.
Where to go next
mcp.run alternative — FAQ
Is mcp.run a directory?
It's more a platform for hosting and running MCP servers (“servlets”) than a browse-and-copy directory. mcptrove is the directory side: curated servers with copy-paste config, capability pages and comparisons.
mcp.run alternative for just finding and running servers locally?
mcptrove lets you find a server and paste its config to run locally — no account, no hosted runtime, no SDK. You give up mcp.run's hosting and secrets management in exchange for control and zero setup overhead.