Request Early Access Getting Started Guide

Getting Started

From zero to your first running agent in less than 10 minutes.


Join the conversation

We’re building MCProspero in the open and we genuinely want your input. Found a bug? Have an idea for a feature? Built a cool agent you want to show off? Head over to our feedback repo:

Or email us at hello@mcprospero.ai if that’s more your style.


0. Caveats

Please know that the early-access program is for people who want to help make MCProspero great, and are willing to endure a few hiccups along the way. We will want to get your feedback and ideas and learn what works and what needs fixin’. But PLEASE do not use this early access platform for business-critical agents, and definitely not for life-critical agents.


1. What you need

Here’s the short list:

MCProspero provides everything else – AI tokens, infrastructure, scheduling, tool integrations. No API keys required to get started.


2. Connect MCProspero to your MCP client

This is step one. You need to add MCProspero as an MCP server in your MCP client. The server URL is always:

https://api.mcprospero.ai/mcp

If you haven’t been approved for early access yet, you’ll see an early access page when you try to authenticate. Request access here and we’ll add you as quickly as we can.

Claude Desktop (macOS & Windows)

  1. Open Claude Desktop, click the Settings gear icon, then Connectors
  2. Click Add Custom Connector
  3. Enter the server URL: https://api.mcprospero.ai/mcp
  4. Name it MCProspero and click Save
  5. You’ll be asked to sign in – pick Google, GitHub, or email magic link
  6. First time only: a quick welcome screen asks for your name and timezone (~30 seconds)

Once connected, start a new conversation and Claude will know about your agents.

claude.ai (Web)

  1. Go to claude.ai
  2. Click your profile icon (bottom-left), then Settings, then Connectors
  3. Click Add Custom Connector
  4. Enter the server URL: https://api.mcprospero.ai/mcp
  5. Name it MCProspero, save, and authenticate when prompted

Claude Mobile (iOS & Android)

  1. Open the Claude app, tap your profile icon, then Settings, then Connectors
  2. Tap Add Custom Connector
  3. Enter the server URL: https://api.mcprospero.ai/mcp
  4. Authenticate when prompted

Your agents work from any device. Create on desktop, check status from your phone.

ChatGPT

  1. Open ChatGPT, go to Settings, then Apps
  2. Add a new MCP server with the URL: https://api.mcprospero.ai/mcp
  3. Authenticate when prompted

Cursor

Add MCProspero to your .cursor/mcp.json:

{
  "mcpServers": {
    "mcprospero": {
      "url": "https://api.mcprospero.ai/mcp"
    }
  }
}

Then authenticate when prompted.

Other MCP clients

MCProspero works with any MCP client that supports remote servers. The server URL is always https://api.mcprospero.ai/mcp. Authentication uses standard OAuth 2.0.


3. The dashboard

In addition to conversation, MCProspero has a web dashboard at app.mcprospero.ai. You can sign in there with the same account you used to connect your MCP client.

The dashboard is your at-a-glance view:

Your workflow


4. Connect your services

Before your agents can read email, send Slack messages, or check your calendar, you need to connect those services. This is a one-time setup – after that, all your agents can use them.

You can connect services two ways: through conversation (ask Claude) or from the Settings page on the dashboard. Both do the same thing.

Gmail & Google Calendar

“Connect my Gmail”

Claude will give you an OAuth link. Click it, sign in with your Google account, and approve the permissions. This connects both Gmail and Google Calendar in one step.

Important – we haven’t been officially approved by Google yet (it’s in the works), so you may see a warning page asking you to confirm you trust the app. That’s temporary and expected.

If you have more than one Gmail account, use labels to keep them straight:

“Connect my work email”

“Connect my personal email”

You can disconnect anytime by saying “Disconnect my Google account” or from the Settings page on the dashboard.

Outlook & Office 365 Calendar

“Connect my Outlook”

Same OAuth flow as Google – click the link, sign in with your Microsoft account, approve permissions. This connects both Outlook email and Office 365 Calendar.

Works with personal Microsoft accounts (outlook.com, hotmail.com) and work/school accounts (Microsoft 365).

If you have both Google and Microsoft, labels keep things organized:

“Connect my personal email” (Google) and “Connect my work email” (Microsoft)

Your agents know which account to use based on the label.

Slack

“Connect my Slack”

You’ll get an OAuth link to authorize MCProspero in your Slack workspace. Once connected, your agents can send messages and reply to threads, read channel history, list channels and users, and add reactions.

Some companies restrict adding custom apps to Slack – you may need to check with your IT team first.

GitHub

“Connect my GitHub”

Authorize via OAuth, and your agents can list repositories, search and create issues, add comments and labels, and read pull requests.

GitHub also supports webhook triggers – you can configure an agent to respond to events like deployments, new issues, or pull requests. More on webhooks in section 10.

HTTP API credentials

If your agent needs to call an API that requires authentication (weather services, data providers, internal APIs), MCProspero can store the credentials securely and inject them into requests automatically. Your API keys never appear in the conversation.

Tell Claude something like:

“I need to connect to the OpenWeatherMap API. I have an API key.”

Claude will give you a secure browser link. Paste your API key in the form – you don’t need to worry about how it gets injected. Claude figures that out from your agent’s system prompt.

Here’s what the platform supports behind the scenes:

Method What you enter How the platform uses it
API key as query param Your API key (e.g. abc123def) Appended to the URL: ?appid=abc123def
Bearer token Just the token (e.g. sk-abc123) Sent as header: Authorization: Bearer sk-abc123
API key as header Your API key (e.g. abc123def) Sent as header: X-API-Key: abc123def
Basic auth username:password Base64-encoded in header: Authorization: Basic ...

To rotate a credential, disconnect it and reconnect with the new key.

Phone number and SMS

Want text message alerts? Connect your phone:

“Connect my cell phone to MCProspero”

Claude will give you a browser link where you’ll consent to receive SMS, enter your number, and verify with a code. Once connected, any agent can text you:

“Text me when my deploy agent detects a failure”

Check your connections

Ask Claude:

“What services do I have connected?”

Or just open the Settings page on the dashboard – your connections are listed right there.

Understanding accounts

When you say “connect my work Gmail,” MCProspero creates an account called “work.” All services you connect with the same label get grouped into that account. This is handy for keeping things organized:

“Check my work email, DM me at my work Slack…”

Your default account is called “personal.” When you say “Slack me the update” without specifying, that’s where it goes. The email you signed up with is automatically part of your personal account.

You can also use accounts to represent people you interact with:

“Create an account called wife, email honey@example.com”

Then later:

“Forward this alert to my wife”

To see all your accounts, say “List my accounts” or check the Settings page on the dashboard. You can manage accounts from either place.

How email works

All emails sent by your agents go through the MCProspero platform – not directly from your Gmail or Outlook. This means every email gets content scanning, abuse controls, and consistent formatting.

Here’s what happens:

You don’t need to think about any of this when creating agents. Just say “email me a summary” or “send the report to my boss” and the platform handles the rest.


5. Build your first agent

Just describe what you want. Claude is your design partner – it’ll ask clarifying questions, suggest the right tools, and help you get the details right.

Try something like:

“Can you help me make an agent to check Hacker News every morning at 8 AM and email me a summary of the top AI stories?”

What happens next

Claude will walk you through it:

  1. Create – Claude asks clarifying questions (what time zone? how many stories? what email?) and builds the agent.
  2. Dry run – MCProspero runs a test with no real external calls to figure out what tools and endpoints the agent will need.
  3. Manifest review – You get a manifest listing everything the agent plans to do – tools, domains, schedule. This is how MCProspero keeps you in control.
  4. Approve – You review and approve the manifest. The agent starts running.
  5. Runtime enforcement – Only what you approved is allowed. If the agent tries something new, it stops and asks for re-approval.

Schedules

Agents support both simple intervals and cron expressions:

Agents can also wake up on webhook triggers. If you find yourself wanting to run every minute polling something, a webhook is probably what you want instead.

Managing your agents

You have two ways to manage agents: conversation and dashboard.

In conversation, just ask:

“Show me my agents”

“How is my hacker news agent doing?”

“Stop the weather agent”

“Run the email agent now”

“Change the HN agent to run at 7 AM instead”

“Have the HN agent also mail my sister”

On the dashboard, the fleet view shows all your agents at a glance. Click an agent to see details, recent runs, and logs. Use the buttons to start, stop, run now, approve, or delete – no conversation needed for quick actions.

If something goes wrong

Ask Claude. Seriously, that’s the whole answer.

“Why didn’t I get the email from my weather agent?”

“My HN agent is in ERROR state. What happened?”

Claude can read the logs, spot the issue, and help you fix it – often in one exchange.

Common causes of ERROR state:

Either way, Claude will help you update the agent and re-approve.


6. Example agents

Here are some agents running on MCProspero today. Use these as inspiration – or just paste them into your conversation.

Weather & Sports

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

Morning News Digest

“Check Hacker News every morning at 8 AM and email me a summary of the top AI and startup stories. Keep it to 5-7 items with a few sentence summary.”

Hollywood Trades Digest

“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.”

Meeting Prep

“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.”

Deploy Digest

“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.”

Urgent Email Monitor

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

Outlook Inbox Summary

“Every morning, check my Outlook for unread messages from my boss and summarize them in Slack.”

Watchdog Agent

“Check every hour if my weather agent is actually sending emails. If I haven’t gotten one in 8 hours, Slack me.”


7. Optimize your agents

Your agents will work out of the box, but Claude can help you make them better.

Ask Claude to review your agent

“Look at my HN agent’s recent runs. Is it working well? Any suggestions?”

Claude can read the agent’s logs and transcripts, spot issues, and suggest improvements – like better search queries, smarter filtering, or more efficient tool usage.

Use checkpoints for memory

Agents can save state between runs using checkpoints. This is powerful for things like:

Just describe the behavior you want and Claude will set up the checkpoints automatically.

Tune your agents after they’ve run a few times. A small tweak can dramatically reduce the number of AI tokens your agent needs – which matters because the platform gives you a fixed daily budget.

PII detection settings

MCProspero scans tool inputs and outputs for personally identifiable information (SSNs, credit cards, phone numbers). You can configure how you want the system to respond:

Set your overall preference in account settings (through conversation or the Settings page), and override it per connection on a given agent when needed.

Notifications

Configure how you want to be notified about system events – PII detections, budget exhaustion, agent errors. Start by saying:

“Show me my account settings”

Then tell Claude what you want to change:

“Text me on PII warnings”

“Slack me when an agent hits an error”


8. Choosing a model

MCProspero uses Claude Haiku 4.5 by default – it’s fast, affordable, and great for most agent tasks. But you can upgrade to more powerful models when you need deeper reasoning.

Model Tier Best for Cost
Haiku 4.5 Standard Monitoring, alerts, email triage, data pulls 1x
Sonnet 4.6 Advanced Classification, summarization, analysis 3.75x
Opus 4.6 Deep Research, complex reasoning, nuanced writing 18.75x

To use a different model, tell Claude when creating or updating an agent:

“Create this agent using Sonnet”

“Switch my competitor watch agent to Opus – I want deeper analysis”

“Use the advanced model for this one”

You can reference models by name (Haiku, Sonnet, Opus) or tier (standard, advanced, deep). More powerful models use more of your daily budget, so use them where the quality difference matters.

Bring your own API key

Want to use your own LLM provider API key (Anthropic, OpenAI, Groq, etc.)? Tell Claude:

“I want to use my own Anthropic API key”

Claude will give you a secure browser link to enter your key – it never appears in the conversation. Once stored, your agents use your key instead of the platform’s. Great for heavier workloads or if you want to use a specific provider.


9. Webhooks

In addition to schedules, agents can be triggered by webhooks – external events that wake your agent up on demand.

Tell Claude:

“Set up a webhook trigger for this agent”

You’ll get a unique webhook URL and signing secret. Point your external service (GitHub, Stripe, your own app) at that URL, and your agent runs whenever the webhook fires.

Agents can use both schedule and webhook triggers simultaneously – for example, a daily summary on schedule plus immediate runs when a GitHub issue is created.

Webhook payloads are validated (HMAC-SHA256 signature), size-capped, and scanned for sensitive data before being passed to your agent.

Important: You’ll be asked to approve the agent’s manifest before webhooks will work. If you configure the external service before approving, it may report delivery errors until the agent is approved and running.


10. Usage & limits

During the preview, MCProspero provides platform-hosted AI tokens at no cost. To keep things fair, there are daily limits:

Limit Preview
Agents Up to 10
Daily budget 100,000 cost units (resets at midnight UTC)
Tool calls per run 20
Platform email sends 10 per agent per hour
SMS sends 10 per agent per hour

What are cost units? One cost unit equals one Haiku token. If you use Sonnet (3.75x) or Opus (18.75x), each token costs proportionally more units.

Check your usage anytime:

“What’s my budget status?”

Or glance at the status bar on the dashboard – it shows your daily usage at a glance.

If you hit your daily limit, agents will pause and resume the next day. No data is lost. You can also bring your own API key for higher limits – see Choosing a model.


11. Runtime constraints

Your agents run in a managed environment with a few guardrails worth knowing about:

If your agent hits any of these limits, ask Claude what happened – it can read the logs and help you adjust.


Good luck!

Ready to build something? Open Claude, connect MCProspero, and describe your first agent. Claude will take it from there.

Or open the dashboard and take a look around.

Questions? Ideas? Join the conversation or email hello@mcprospero.ai.

Request Early Access