Data Parrot

npx add-mcp 'https://api-v3.dataparrot.ai/api/v3/data-parrot/mcp'

Data Parrot MCP for HubSpot

MCP changelog

Data Parrot MCP brings AI revenue analysis of your HubSpot data into your AI tools so you can review your forecast, investigate at-risk deals, and understand pipeline changes through conversation.

Ask about sales activity, customer health, and why deals are won or lost. Follow up to examine individual deals, supporting activity, and the reasoning behind Data Parrot’s assessments.

Data Parrot connects to HubSpot and learns how your company sells. Your AI tool uses analysis Data Parrot has already completed, together with supporting CRM data, to answer your questions.

Get a Data Parrot account · Visit Data Parrot · MCP documentation · Connection guide · Agent installation guide · Prompt library

What is the Data Parrot MCP server?

Data Parrot is an AI revenue intelligence platform for CEOs, CROs, and revenue leaders using HubSpot. It monitors every deal, pipeline movement, and customer interaction so revenue teams can trust the forecast, catch slippage early, and focus the team where it matters.

The Data Parrot MCP server brings that analysis into AI tools that support the Model Context Protocol. Instead of starting with raw CRM records and rebuilding the analysis in every conversation, your AI tool can use Data Parrot's deal health, forecast risk, pipeline movement, customer health, sales activity, and win/loss analysis.

This repository contains connection instructions and a Gemini CLI extension. The MCP server itself is hosted by Data Parrot at https://api-v3.dataparrot.ai/api/v3/data-parrot/mcp. There is no local server to run.

Why use Data Parrot with HubSpot?

CRM has the data. Data Parrot AI gives you the truth. Data Parrot has already analyzed how your company sells and what is happening across the sales organization, from individual deals and customers to the forecast and the reasons behind wins and losses.

Get answers to your revenue questions without building reports or reading through emails and meeting notes yourself. Your CRM doesn't have to be perfect first.

Every conversation starts with analysis Data Parrot has already done. You can ask what changed, what is at risk, and where the team needs to focus without first explaining your pipeline or asking the model to reconstruct deal history record by record.

What can Data Parrot do through MCP?

Sales forecasting

Build a realistic forecast from closed-won revenue, the remaining pipeline, deal-level risk, best-case upside, and the deals most likely to move the number. You can ask where the business will land this month or quarter and inspect the deals behind the answer.

Deal health and forecast risk

Review purchase intent, deal progress, stage confidence, close-date confidence, buyer engagement, and forecast risk. Find high-value deals that are stuck, going quiet, slipping, or showing stronger intent than their stage suggests.

Pipeline analysis

See how the pipeline changed over a period and which deals caused the movement. Data Parrot can help explain pipeline growth, shrinkage, pushed close dates, stage changes, and the deals that need attention now.

Sales activity and coaching

Use activity and deal evidence to identify where reps need coaching and which deals belong in the next review. Go into coaching conversations knowing where buyer engagement or deal progress needs attention.

Customer health

Find high-value customers that need attention and review the supporting account activity. This helps teams spot customer risk before the next renewal or executive review.

Win/loss analysis

Data Parrot analyzes why deals are won or lost so you can see what to repeat and what needs to change. Compare the reasons across deals, segments, owners, competitors, and time periods without manually reading every record.

Questions you can ask

Start with questions your revenue team already asks:

• "Where will we land this month? Include closed-won revenue, a realistic remaining forecast, best case, and the deals that could move the number."
• "Which deals put this month's forecast at risk?"
• "Which high-value deals have strong buyer intent but are stuck or going quiet?"
• "What grew or shrank our pipeline this quarter, and which deals caused the change?"
• "Which reps need coaching right now, and which deals should we review with them?"
• "Which high-value customers should we be worried about?"
• "Why are we winning and losing?"

See the Data Parrot prompt library for more examples covering forecasts, open deals, pipeline movement, sales activity, customer health, and win/loss analysis.

Use the analysis in your existing workflow

The answer does not have to stay in chat. When your AI tool has access to the right file or productivity apps, it can use Data Parrot's analysis to help create a forecast deck, prepare a deal review, update a planning document, or build a working spreadsheet.

Data Parrot supplies the revenue analysis and supporting details. Your connected AI tool handles the document or workflow in the apps you have authorized.

Data Parrot MCP vs. HubSpot MCP

Data Parrot MCP and HubSpot MCP solve different jobs and can be used side by side.

	Data Parrot MCP	HubSpot MCP
Primary job	Answer revenue questions using analysis Data Parrot has already produced	Give an AI tool access to supported HubSpot CRM records and activities
Best for	Forecasting, deal risk, pipeline movement, coaching, customer health, and win/loss analysis	Looking up CRM data and taking supported actions in HubSpot
Information returned	Data Parrot analysis with supporting HubSpot details	Supported HubSpot records, activities, and account information
Revenue data access	Read-only	Read and write access depends on the supported object and configured scopes

Use HubSpot's MCP server documentation for direct CRM access. Use Data Parrot when the question requires the sales analysis behind the records.

How it works

1. Data Parrot connects to HubSpot and analyzes deals, pipeline movement, sales activity, customer accounts, and won and lost business.
2. You add the Data Parrot connection to your AI client using its setup guide below.
3. Data Parrot asks you to sign in through your browser and choose the account you want to use.
4. Ask about your forecast, deals, or pipeline. Your AI tool uses Data Parrot's existing analysis and supporting details to answer.

Before connecting

You need a paid Data Parrot account on Pro or Max, with HubSpot connected. MCP access is included at no additional Data Parrot charge.

New to Data Parrot? Get a Data Parrot account.

You also need access to the workspace with MCP enabled for your user. Your organization may need to allow the connection in your AI client. See the plans and permissions FAQ.

Setting	Value
Server name	data-parrot
Endpoint	https://api-v3.dataparrot.ai/api/v3/data-parrot/mcp
Transport	Streamable HTTP
Authentication	OAuth through your browser, discovered by the client

The normal connection flow requires no API key, HubSpot private-app token, or pasted bearer token. Preserve existing client settings and reuse an existing Data Parrot connection for this endpoint instead of creating a duplicate.

Connect your AI client

Choose your client and follow its setup guide. Client versions, account plans, and organization policies can affect the available connection controls.

AI client	Setup summary	Instructions
Claude.ai / Claude Desktop	Add a custom connector, then sign in and enable it in your conversation.	Claude connector guide
Claude Code	Add the HTTP server from your terminal, then authenticate in Claude Code.	Claude Code guide
ChatGPT	Add the remote connection through Developer mode and choose OAuth.	ChatGPT guide
Codex app	Add a Streamable HTTP server in MCP settings, then authenticate.	Codex app guide
Codex CLI	Add the remote server and complete browser sign-in; reuse the app connection if already configured.	Codex CLI guide
Cursor	Install the separate Data Parrot plugin from GitHub, then authenticate.	Cursor guide
Grok.com	Add a custom connector from the Connectors page, then sign in.	Grok.com guide
Grok CLI	Add the HTTP server, then authenticate from the MCP Servers menu.	Grok CLI guide
Cline	Add a remote Streamable HTTP server and complete browser authentication.	Cline installation instructions
Gemini CLI	Install this extension or configure the HTTP endpoint directly, then authenticate.	Gemini CLI installation instructions

For other clients, see the custom connection guide. The Cursor/Grok Bot package remains separate; the Grok.com and Grok CLI guides above describe different clients from Grok Bot.

Permissions and verification

Revenue-data access is read-only. Data Parrot MCP cannot edit revenue records in Data Parrot or HubSpot.

Connection setup has separate effects. Selecting or switching a workspace changes the connection binding. Choose the workspace yourself.

Verify the connection

Check that Data Parrot tools are available. If workspace access is unclear, ask the client to call get_data_parrot_access_status. Then ask:

Where will we land this month?

Confirm that the client actually called a Data Parrot business tool successfully and used the intended workspace and period. A chat answer alone is not proof of a connection. If the workspace has no applicable data, the answer should say so. Review important conclusions before acting on them.

If sign-in succeeds but revenue tools are unavailable, check the returned access status: membership, paid-plan eligibility, the HubSpot connection, MCP access for your user, or workspace selection may need attention. Follow the returned setup guidance. If membership is missing, ask your Data Parrot admin to invite you.

Validation status: the installation instructions have been checked against provider documentation. Client installation, authenticated OAuth, workspace access, token refresh, and successful business-tool execution have not been tested as part of this package review. See the installer guide for the documentation-check date and version details.

Support

For connection problems, use your client's reconnect/authentication controls and review the MCP FAQ. A browser GET to the MCP endpoint can return HTTP 405; that is not a complete MCP health check.

Follow your client's linked guide to remove the connection. Cline and Gemini removal instructions are in the installer guide.

Contact support@dataparrot.ai with the client version and error message, excluding credentials and customer data.

License

The files distributed in this repository are available under the MIT License. This repository does not distribute the hosted Data Parrot service or its private server implementation.