FAQ

Meet Greg Warden, a software engineer and product builder based in New Jersey. What started as a learning exercise — "how would one think about data privacy in an agentic world" — turned into "let's build some agents and find the pointy bits," then "there has to be a better way to build and run agents," then "I want this on my phone and my computer," then "wow I want this thing to be real," then "I am sure I am not alone." And here we are. Easy to build, secure, accessible... useful AI agents for everyone — not just developers.

Got questions? hello@mcprospero.ai

MCProspero is an MCP Server that turns your conversations with Claude — an MCP Client — into AI-powered agents that work for you around the clock. You authenticate tools (like Gmail, Slack) with MCProspero, and through the magic of the MCP protocol, Claude learns about those tools and becomes your Agentic Design Partner — your words, Claude's words, your brain, Claude's brain. Once you and Claude figure out what you want, and the best way to do it, MCProspero takes care of the rest.

MCProspero is deployed on secure, robust infrastructure, accessible from anywhere, that you don't need to worry about. Launch your agents and we keep them running on whatever schedule you want. "Check Hacker News every morning for new AI stories and send me a distilled version of the important ones. And also, remind me to bring an umbrella if it's going to rain."

MCP stands for Model Context Protocol — an open standard that lets AI assistants like Claude connect to external tools and services. MCProspero is an MCP server, which means any MCP-compatible client (Claude Desktop, claude.ai, ChatGPT, and others) can connect to it and unlock the power of MCProspero.

No. You create agents by describing what you want in a conversation. Claude figures out the tools, the schedule, and the edge cases. If something breaks, Claude helps you diagnose and fix it — no logs, no config files.

Well... Combine your information, with information from public sources accessible through the web, with information the world's best LLMs already know, and then ask those same LLMs to help you make sense of it. Really, the limit is your imagination.

That said, some of the cool information sources out there are private, and require MCProspero to do light integration work for you to make full use of it. So you can't do everything you can think of. But the list of what you can do is growing every day.

Some examples of things running on the platform today:

"Every morning, send me today's weather for Ridgewood, NJ and tell me which Boston sports teams are playing."

"Check Hacker News every morning at 8 AM and email me a summary of the top AI and startup stories."

"Every morning at 8 AM Pacific, pull from Hollywood Reporter, Variety, Deadline, and IndieWire and send me a digest of new films going into production, getting greenlit, or getting major casting attached."

"Every morning, look at my calendar and see if there's any recent email from anyone I'm meeting with that I should read first."

"When a GitHub deploy finishes, summarize the commits for the #dev channel in Slack, then write a customer-friendly version and email it to marketing."

"Every hour, check my Gmail for unread messages from my wife. If anything looks urgent, Slack me immediately."

Agents can connect to Gmail, Slack, Calendar, GitHub, HTTP APIs, and more. New integrations are added regularly.

MCProspero supports any OpenAI-compatible model. By default, agents select one of the Anthropic models (Haiku, Sonnet, Opus) running on AWS Bedrock to orchestrate the agent runtime. You can also use other AWS Bedrock models from OpenAI, Meta (Llama), Mistral too, if you want. And in our Builder tier, you can connect any remote OpenAI-compatible provider/model.

No. Use any MCP client to manage your agents. Any OpenAI-compatible model can serve as your agent's brain — Claude, OpenAI, Llama, Mistral, or whatever comes next.

Open standards on both sides. No lock-in, ever.

MCProspero is currently in invite-only preview and free to use during the preview period. No API key required — agents run on platform-provided AI tokens.

After the preview, we plan to offer a generous free tier and monthly subscriptions for more capacity and advanced data security and enterprise features.

Request early access

Yes — and not because we asked the AI nicely. MCProspero enforces safety architecturally:

During agent creation, MCProspero surfaces what tools and endpoints the agent wants to use. You approve the list. At runtime, anything not on the approved list is blocked.

You can also opt in for PII security scanning. If data starts leaking, the agent gets automatically suspended.

LLM traffic runs on AWS Bedrock and never touches the public internet. OAuth tokens and API keys for your tool integrations are encrypted at rest using AWS KMS.

Unleash the power of the best LLMs without worrying about hallucinations.

MCProspero was designed with enterprise in mind. Although the initial launch is geared towards individuals and small teams, we can support a variety of models: SaaS shared, SaaS dedicated, hybrid (SaaS control plane, on-prem agents), and full on-prem. Same code, different config. Soon we'll be looking for a few enterprise design partners. Interested? hello@mcprospero.ai

Yes. MCProspero works from any MCP client — Claude Desktop, claude.ai on the web, or Claude mobile, just to name a few. Create an agent on your desktop, check its status from your phone. Or vice versa. Your agents go where you go. Greg tends to wake up in the morning with a new agent idea — opens Claude mobile, and speaks it into existence (yes, the speech-to-text on his phone is pretty good. Why type when your hands should be holding a cup of coffee?).

We don't have "open reqs," but if you are incredibly passionate about our mission, tell us how you want to help. Especially in areas Greg and Claude Code are bad at (which is not a small list, actually). hello@mcprospero.ai

Prospero is the sorcerer in Shakespeare's The Tempest.

You need a sorcerer.