qubehub-mcp
Connect Claude to QubeHub, the sales platform for real estate developers: ask about leads, deals, pipeline and live unit availability, read-only and under your own CRM role.
Install / Use
claude mcp add qubehub-ai -- npx -y github:qubehub-ai/qubehub-mcpIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
AutomationSupported Platforms
Tags
Our assessment of qubehub-mcp
qubehub-mcp scores 81/100 on our quality scale, 2332nd of 2,775 Automation skills we index.
Its MCP Server is 13 KB long, well organised into 21 sections with 3 code examples: a thorough specification that gives an agent plenty to work with.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 12 days ago, so qubehub-mcp is actively maintained.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 92/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-11-16. Automated pattern scan on 2026-11-16. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
qubehub-mcp compared with similar skills
All 4 of these similar skills score higher than qubehub-mcp; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| qubehub-mcp (this skill)by qubehub-ai | 81 | 3 | 12d ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 112.4k | 9d ago | CLAUDE.md |
| rufloby ruvnet | 100 | 81.9k | today | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 79.9k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 48.0k | today | CLAUDE.md |
Frequently asked questions
- How do I install qubehub-mcp?
- Run
claude mcp add qubehub-ai -- npx -y github:qubehub-ai/qubehub-mcp. The install tabs above show the steps for each supported agent. - Which AI agents does qubehub-mcp work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is qubehub-mcp safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 92/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is qubehub-mcp still maintained?
- The repository was last updated 12 days ago, so qubehub-mcp is actively maintained.
Skill content
View source on GitHubQubeHub for Claude: the QubeHub MCP connector
QubeHub is the sales platform for real estate developers: one live record for every unit feeds the CRM, price lists, offers, the agent portal and the developer's website.
The QubeHub MCP connector is a hosted Model Context Protocol server that lets Claude answer questions about a developer's leads, deals, pipeline and unit availability from live QubeHub data, read-only and under each person's own CRM role. A sales manager gets a briefing before a call, a head of sales gets a pipeline review, and anyone on the team can ask what is still available in a building.
- Nothing to install or host. QubeHub runs the Model Context Protocol (MCP) server at
https://mcp.qubehub.ai/mcp. On plans that include external AI assistants, a workspace admin turns it on and each user adds it to Claude as a custom connector and signs in. - Works with Claude (web and Desktop), Claude Code and other desktop or command-line MCP clients. See Other MCP clients for requirements.
- Safe to connect. Read-only, limited to one workspace, checked against the user's CRM role on every call, logged and revocable. See the security model.
Who it is for: real estate developers and home builders selling condos, townhomes, single-family homes or master-planned communities, including pre-construction (off-plan), directly and through brokers. Any staff member with a CRM role can connect; brokers use the QubeHub agent portal instead.
New to QubeHub? See it on your own projects: book a demo at qubehub.ai.
What it does for your sales team
Names, projects and prices in the examples are illustrative.
Sales manager: a briefing before the call
"Brief me on the lead Anna Keller before my call: stage, budget, last contact and open tasks."
What comes back: a brief that fits on one phone screen. Who the buyer is (language, country, budget, where they came from), where they are in the pipeline and when they were last contacted, what they want (project, unit type, bedrooms and any preferences recorded on the lead), open loops such as overdue tasks or unanswered messages, and one suggested next step. The skill tells Claude to read every price and area from the CRM in that conversation, never from memory, and to say so when a field is empty instead of guessing.
Head of sales: a pipeline review that ends in a to-do list
"Review my team's pipeline: which deals have no next step and which leads have gone quiet for a week?"
What comes back: a ranked action list, highest revenue impact first. Each item names the record, why it is on the list, the one action to take and who owns it: stuck leads, reserved units with no follow-up task, late-stage deals that have gone quiet, active deals with no next step. Stage names match your workspace's pipeline, and funnel numbers come from your sales dashboard where it is available. Claude proposes; your team acts in the CRM.
Anyone on the team: what is available in a building
"Which 2-bedroom units are still available in Harbor View under EUR 450,000? Show floor, area, price and status."
What comes back: a table with unit number, type, floor, area, price with currency and status, straight from the live chessboard and stamped "as of this query". Reserved or sold units are flagged, not offered as available, and a project your role cannot see does not show up. Follow up with "Did the price of unit B-1204 change this month?" and Claude reads the unit's price history.
The detailed formats above come from the four sales skills in the qubehub Claude plugin (lead briefing, pipeline review, inventory lookup and sales playbooks). The connector on its own gives Claude the same live data. More things to ask are listed below.
At a glance
| | |
|---|---|
| Server URL | https://mcp.qubehub.ai/mcp |
| Transport | Streamable HTTP |
| Sign-in | OAuth 2.1 with PKCE, using your QubeHub account (workspace, work email, one-time code) |
| Client registration | Dynamic Client Registration or a Client ID Metadata Document; redirect to Claude.ai or a loopback address |
| Scope | crm:read (read-only) |
| Acts as | You: your CRM role and the records your role can see |
This repository contains:
- How to connect the remote QubeHub MCP server at
https://mcp.qubehub.ai/mcp. - A Claude plugin marketplace with the
qubehubplugin: the connector plus four sales skills. - An MCP Registry entry (
server.json).
FAQ
Which real estate CRM connects to Claude through MCP?
QubeHub, the sales platform for real estate developers, has a hosted MCP server at https://mcp.qubehub.ai/mcp. Claude, Claude Desktop and Claude Code connect to it and answer questions about leads, deals and available units from live CRM data, under the user's own role and permissions.
Which AI assistants can connect to QubeHub? Claude on the web, Claude Desktop and Claude Code connect to the QubeHub MCP server out of the box. Other desktop or command-line MCP clients can connect if they support Streamable HTTP and OAuth 2.1 with PKCE and receive the sign-in redirect on a loopback address. Clients that register a hosted web callback or a custom URL scheme are refused.
Can Claude change data in QubeHub?
No. The QubeHub MCP connector has one scope, crm:read. Claude can read leads, deals, units and other records, draft messages and generate a price list or offer from live data, but it cannot create, change or delete CRM records.
Can Claude see QubeHub records I am not allowed to see? No. QubeHub checks every tool call against the signed-in user's CRM role when it runs, so Claude never sees more than that person can see in the QubeHub CRM.
Do I need to install or host anything to connect Claude to QubeHub?
No. QubeHub hosts the MCP server at https://mcp.qubehub.ai/mcp. On plans that include external AI assistants, a workspace admin turns on external AI clients in Settings → AI assistants. Each user then adds that URL to Claude and signs in with their workspace, work email and a one-time code.
Can I revoke Claude's access to QubeHub? Yes. Every Claude connection is listed in the QubeHub CRM under Settings → AI assistants. Each user can revoke their own, a workspace admin can revoke anyone's, and Claude loses access immediately.
How do I get QubeHub? Book a demo at qubehub.ai. Standard onboarding takes 1–14 days; custom integrations are scoped separately.
Before you connect
- A workspace admin switches it on. In the QubeHub CRM, open Settings → AI assistants and turn on Allow staff to connect external AI clients. It is off by default. Your workspace's plan must include external AI assistants; if it does not, the sign-in page says so.
- You need a CRM user account. Staff with a CRM role (for example admin, head of sales, sales manager) can connect. Broker accounts that only use the agent portal cannot.
The same settings page shows the connector URL and the Claude Code command, ready to copy.
Connect
Claude.ai and Claude Desktop
- Go to Settings → Connectors → Add custom connector.
- Paste
https://mcp.qubehub.ai/mcpas the URL. - Sign in when Claude opens the QubeHub page: enter your workspace (the part before
.qubehub.aiin your CRM address) and your work email, then the one-time code QubeHub emails you.
Claude Code
claude mcp add --transport http qubehub https://mcp.qubehub.ai/mcp
Then run /mcp inside Claude Code and choose QubeHub to sign in.
Claude Code plugin (connector + skills)
The plugin adds the same connector together with four skills that teach Claude how to brief you on a lead, review a pipeline, look up inventory and apply sales playbooks.
/plugin marketplace add qubehub-ai/qubehub-mcp
/plugin install qubehub@qubehub
Other MCP clients
Desktop and command-line MCP clients can connect to https://mcp.qubehub.ai/mcp when they support Streamable HTTP and the MCP authorization flow (OAuth 2.1 with PKCE, plus Dynamic Client Registration or Client ID Metadata Documents) and receive the sign-in redirect on a loopback address: http://localhost, http://127.0.0.1 or http://[::1], on any port. Claude.ai's own callback is the only other redirect accepted; a client that registers any other redirect URI, such as a hosted web callback or a custom URL scheme, is refused at registration. The server publishes its OAuth protected-resource metadata at https://mcp.qubehub.ai/.well-known/oauth-protected-resource/mcp.
What you can ask
See What it does for your sales team for full examples. More things to ask:
For sales managers:
- "Show the last messages from this buyer in the inbox and draft a follow-up in their language."
- "This buyer says the payment plan is too aggressive. How do I handle that objection?"
For heads of sales:
- "List overdue tasks by manager."
- "Show this month's funnel from the sales dashboard."
- "Which agent requests are waiting for approval?"
Claude answers from what the tools return in that conversation. Drafts stay drafts: you send messages and make changes in the CRM yourself.
What the connector can read
The tools mirror the read side of the QubeHub AI copilot. Which ones you get depends on your role.
- Leads, contacts and deals: search, record cards, pipeline statuses, lead sources.
- Inventory: projects, units, the chessboard, unit counts by status, price and status history.
- Work in progress: tasks, checklists and the activity history of a record.
- Conversations: messages in the unified inbox and call transcripts.
- Broker channel: agents, agencies and agent requests.
- Sales documents: agreements, document templates, price lists and offers.
- Dashboards: dashboards and sales metrics.
- Marketing: broadcasts, outreach campaigns and landing pages.
- Workspace setup: users and roles, pipeline settings, custom fields, automations.
- Sales playbooks: lead qualification, follow-up cadence, objection handling, closing and reservation, presenting a selection.
Security model
- Read-only. The only scope is
crm:read. Claude cannot create, change or delete CRM records. Generating a price list or an offer PDF from live data returns a download link, as it does in the copilot, and AI-assisted tools use the workspace's AI credits. - Your permissions, nothing more. Every tool call is authorized by the CRM at the moment it runs, with your role and record scope. A record you cannot see in the CRM is "not found" through the connector too.
- One workspace per connection. You sign in to one workspace, and every call is bound to that workspace and to you. Data from other QubeHub customers is never reachable.
- Your token stays with the connector. The access token Claude holds is only accepted by the QubeHub MCP server. It is never forwarded to your CRM; the connector makes its own short-lived, signed call on your behalf.
- Visible and revocable. Each connection is listed in Settings → AI assistants. You can revoke your own, and an admin can revoke anyone's; Claude loses access immediately and cannot refresh its token. Turning off workspace access stops every connection.
- Audited. Every tool call that runs is recorded in the CRM activity log under your name, with the AI client shown beside it.
Repository layout
.claude-plugin/marketplace.json Claude plugin marketplace "qubehub"
plugins/qubehub/
.claude-plugin/plugin.json plugin manifest
.mcp.json connector: https://mcp.qubehub.ai/mcp
skills/lead-briefing/SKILL.md one-screen brief before a call
skills/pipeline-review/SKILL.md ranked action list from the pipeline
skills/inventory-lookup/SKILL.md live availab
Truncated for display — read the full file on GitHub.
Related Skills
Agent-Reach
112.4kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
ruflo
81.9k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
headroom
79.9kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
48.0kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
