BELVO MCPs
THE OPEN FINANCE LAYER AI AGENTS TRUST
Enable your AI agents to approve loans faster, deliver accurate insights and build faster with user-permissioned financial data.

Discover the MCPs
Pick the server that fits the job
Three specialized MCP servers designed for seamless integration, live financial data access, and portfolio-wide analysis.
Integrate faster with AI assistance
Cut integration time from weeks to minutes. Your AI client reads our documentation and writes the code. It debugs errors automatically while you scaffold a working sandbox call quickly. Just point your client at the server and ask for exactly what you need.
Check it outUse the Developer MCP to scaffold a Belvo income verification call for a Mexican user and show me a sandbox response.
Here's a working example using Belvo's Developer MCP:
# requirements: belvo-python from belvo import Client c = Client(secret_id, secret_pw, "sandbox") r = c.Incomes.create(link="lnk_mx_a1c9…") # → 200 OK { monthly_income: 4500.00, employer_verified: true }
The right endpoint and a working example, so it writes integration code that runs instead of code your engineer has to correct.
Our error reference on demand, including the cause, the fix, and whether the problem is on your side or ours.
Open your AI client settings and add a custom connector
Paste the Belvo Developer MCP server URL
Ask your first question
Open to everyone and no account is required to access the documentation. Works in coding tools: Cursor, VS Code, Claude Code, Codex, and other MCP-compatible clients.
To make live API calls you'll need a Belvo account.
Access users' verified data
Base your decisions on verified records instead of rough estimates. Your agent can instantly pull banking, employment data or fiscal sources. You ask plain-language questions and get definitive answers back. Every pull enforces strict user consent and leaves a complete audit trail.
How much did this user spend on Uber over the last year?
R$ 4.312,80 across 214 rides in the last 12 months.
| Quarter | Rides | Spend |
|---|---|---|
| Q3 2025 | 48 | R$ 962,40 |
| Q2 2025 | 57 | R$ 1.148,10 |
| Q1 2025 | 61 | R$ 1.284,60 |
When does this user receive their salary?
Based on the past year, ACME LTDA most likely pays on day 25 of the month.
Earnings and employer as filed on the source, so you stop underwriting on self reported numbers.
Categorized spending and the recurring bills, so you know what is left before you lend more.
An ownership check and a job stability score before you approve or pay out.
A raise, a layoff, or a new tax filing, picked up as it lands rather than at the next review.
Talk to our team to activate this feature
Add the server URL in your client settings
Your customer approves access through Belvo
Ask questions about that customer
Billed through standard API calls. Talk to our team to start.
Analyze your entire customer base
Move portfolio analysis directly into your AI client. Ask questions about your whole customer base. The agent reads aggregated figures. It never accesses individual customer records.
Analyze the distribution of the credit card limits of my users.
Most limits sit between R$ 2k and R$ 5k. The distribution tails off steadily, with a small segment above R$ 10k.
How limits and exposure are spread across your customer base.
What share of your customers changed income last quarter, and which segment moved.
Which spending categories drove the biggest shift, without exporting to a separate dashboard.
Add the Belvo Analytics server URL
Sign in with your production account
Give your AI access to aggregated portfolio data
Ask portfolio questions in plain language
Requires a production Belvo account.
CONNECT YOUR ACCOUNTS AND TEST IT LIVE
Test the Data Access MCP with our live demo. Connect your personal accounts and ask the AI plain-language questions. You experience the exact consent flow your users will navigate. Verify the data speed and evaluate the accuracy immediately.

FREQUENTLY ASKED QUESTIONS
Your AI agents need verified data. Web scraping is slow and risky. Belvo turns our existing open finance rails into tools your agents can call directly. If your agent cannot reach trustworthy financial data, it stalls.
The MCP servers carry no extra cost. You only pay for the underlying API calls. We use the exact same pricing model as a direct API integration.
The Developer MCP and Data Access MCP are live in Brazil and Mexico. The Analytics MCP is currently on our product roadmap.
The protocol itself is just a basic bridge. Belvo secures the actual data underneath. We enforce strict identity binding and least privilege on every call.
Your agent can access banking and employment sources in Brazil, and employment and fiscal sources in Mexico.
Any MCP-compatible client works perfectly. You can easily connect Claude, ChatGPT, Microsoft Copilot, Gemini, and Cursor.









