> ## Documentation Index
> Fetch the complete documentation index at: https://docs.usequanta.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect Quanta to your AI assistant with MCP

> Connect Quanta to Claude, ChatGPT, or another MCP client with OAuth sign-in to ask read-only questions about your financial statements, spend, and balances.

The Quanta MCP server connects your accounting data to Claude, ChatGPT, and other AI assistants that support the Model Context Protocol (MCP), an open standard for connecting AI assistants to external tools and data. Once connected, you can ask about your financial statements, spend, receivables, payables, and bank balances in plain language, and download Excel reports, without leaving the conversation.

> [!NOTE]
>
> Your books stay read-only. The connector cannot create, edit, or delete accounting records.

## What you can do

### Understand your financials

Pull your income statement, balance sheet, and cash flow statement into the conversation. Compare periods and drill into the activity behind the numbers.

> "Show me our financial statements for the last six complete months and summarize the biggest changes."

### Explore where money went

Break down cash-basis spend by vendor or category. Distinguish money paid out from expenses recognized on your income statement.

> "Which vendors accounted for the most cash-basis spend last quarter?"

### Review receivables and payables

Explore outstanding invoices, unpaid bills, payment status, and aging.

> "Which customer invoices are still outstanding, and how much is overdue?"

### Check your latest bank balances

View the most recent provider-reported balances for connected bank and investment accounts, including how recently they were updated.

> "Show our latest bank balances and flag any accounts with stale balance data."

### Download reports

Ask for an Excel export of your financial statements or your accounts receivable aging report.

> "Export last quarter's financial statements to Excel."

## Before you connect

You need:

* A Quanta account with access to the organization you want to connect.
* MCP enabled for that organization. Contact your Quanta team to request access.
* An AI app that supports remote MCP connections and OAuth sign-in.

> [!NOTE]
>
> For company-managed Claude or ChatGPT accounts, a workspace owner or administrator may need to approve or add the connection first.

## Connect Quanta

Use this server address for every supported client:

```text theme={"system"}
https://mcp.usequanta.com/mcp
```

This is a connection endpoint, not a webpage. Paste it into your AI app's MCP setup, not your browser's address bar.

> [tab] Claude
>
> **For an individual account**
>
> **1. Open connector settings**
>
> Open Claude and go to **Customize → Connectors**.
>
> **2. Add a custom connector**
>
> Click **+ Add → Add custom connector**.
>
> **3. Enter the server details**
>
> Name the connector **Quanta** and enter the server address above. Use OAuth sign-in. You don't need a Quanta API key or manually supplied OAuth credentials.
>
> **4. Sign in to Quanta**
>
> Click **Connect**, sign in to Quanta, and select your organization.
>
> **5. Turn on Quanta in a conversation**
>
> Enable Quanta for your conversation from Claude's connectors menu.
>
> **For a Team or Enterprise account**
>
> Ask your Claude organization owner, or someone with permission to manage connectors, to add Quanta:
>
> **1. Open organization connectors**
>
> Open **Organization settings → Connectors**.
>
> **2. Add a custom web connector**
>
> Select **Add → Custom → Web**.
>
> **3. Enter the server details**
>
> Enter **Quanta** and the server address above, then complete the authentication setup.
>
> Once Quanta has been added, each person connects through **Customize → Connectors** and signs in with their own Quanta account. Adding the connector to Claude does not give teammates access to anyone else's Quanta account.
>
> See Claude's [custom connector guide](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp) for more detail.

> [tab] ChatGPT
>
> Use ChatGPT's custom MCP connection flow:
>
> **1. Open plugins**
>
> Open [ChatGPT Plugins](https://chatgpt.com/plugins).
>
> **2. Add a custom MCP server**
>
> Select **+ → Add custom MCP server**.
>
> **3. Enter the server details**
>
> Enter **Quanta** as the name and paste the server address under **Connection**.
>
> **4. Sign in to Quanta**
>
> Configure OAuth authentication and complete the Quanta sign-in flow when prompted.
>
> **5. Create the plugin**
>
> Review the permissions and risk notice, then select **Create** as a plugin.
>
> **6. Use Quanta in a conversation**
>
> Install the resulting plugin. Start a new conversation, type `@`, and select **Quanta**.
>
> Custom MCP availability and setup controls depend on your account and workspace policies. If the option is missing, ask your workspace administrator or check [OpenAI's custom MCP setup guide](https://developers.openai.com/plugins/deploy/connect-chatgpt).

> [tab] Other MCP clients
>
> Quanta works with clients that support remote Streamable HTTP and OAuth authentication.
>
> | Setting | Value |
> | - | - |
> | Connection name | Quanta |
> | Server URL | `https://mcp.usequanta.com/mcp` |
> | Transport | Streamable HTTP |
> | Authentication | OAuth (browser-based sign-in) |
> | Quanta API key | Not required |
>
> Add a remote MCP server in your client, enter these details, and complete the browser sign-in flow. Select the Quanta organization you want to use, then return to your client to start asking questions.
>
> Configuration formats vary by client. If your client requires manually registered OAuth credentials or does not support browser-based authentication, contact your Quanta team for guidance.

## Available tools

You don't need to call these tools yourself. Your assistant chooses them based on your request.

| Tool | What it does | Changes data |
| - | - | - |
| `list_datasets` | Lists the accounting datasets available to your organization. | No |
| `describe_dataset` | Explains a dataset's fields, structure, and relationships. | No |
| `query` | Runs a validated, read-only query against approved accounting datasets. | No |
| `get_financial_report` | Retrieves income statements, balance sheets, and cash flow statements for the conversation. | No |
| `export_financials` | Generates an Excel export and returns a download link. | Creates an export file |
| `list_recipes` | Finds available accounting playbooks. | No |
| `get_recipe` | Loads guidance for answering a particular type of question. | No |
| `preview_recipe_change` | Lets controllers preview a change to an organization playbook. | Saves a preview only |
| `apply_recipe_change` | Publishes a previewed playbook change after explicit approval. | Updates an organization playbook (controllers only) |
| `give_feedback` | Records feedback about whether an answer addressed your question. | Records feedback |

Available datasets cover trial balances, general ledger activity, invoices, bills, recognized revenue, cash-basis spend, latest bank balances, transaction matches, and your chart of accounts.

### Accounting guidance, built in

Quanta includes recipes: short playbooks that help your assistant choose the right data and accounting approach.

For example, a question about cash-basis spend should not be answered with accrual expenses from the income statement. Recipes help the assistant make that distinction and explain relevant limitations.

Controllers can also create organization-specific playbooks. Changes follow a separate preview-and-approval flow and do not change your accounting records.

## Access and safeguards

* **Individual sign-in.** Each person connects with their own Quanta account. Requests are scoped to the organization selected during sign-in.
* **Read-only accounting access.** The connector cannot post journal entries, edit invoices, approve bills, or otherwise change your books.
* **Separate playbook permissions.** Organization playbook changes are controller-only and require a preview and explicit approval.
* **Bounded queries.** Queries run only against approved datasets, with row limits, time limits, and concurrency controls.
* **Audited activity.** Quanta records MCP tool activity for auditing.

Data returned to your assistant is processed by the AI provider you choose. Review that provider's privacy settings and your organization's policies before connecting. For how Quanta handles your data, see the [Quanta privacy policy](https://www.usequanta.com/privacy-policy).

## Frequently asked questions

> [accordion] Do I need to write SQL?
>
> No. Ask in plain language. Your assistant handles dataset discovery and queries.

> [accordion] Is the data live?
>
> The connector reads the data available in Quanta when it runs. Freshness depends on the underlying integration and accounting status. Provider-reported bank balances may be older than the current day.

> [accordion] Can it change my books?
>
> No. Accounting records remain read-only. Controllers can separately manage organization playbooks through the preview-and-approval flow.

> [accordion] Why does an answer take a moment?
>
> Your assistant may consult a playbook, inspect datasets, and run several queries before responding.

> [accordion] Can I export a full report?
>
> Yes. Ask for an Excel financial statement export or an accounts receivable aging export. Download links expire after 15 minutes, so ask your assistant for a fresh link if one stops working. For larger row-level reports outside those exports, use the Quanta app.

> [accordion] How do I disconnect?
>
> Disconnect or remove Quanta in your AI app's connector or plugin settings.

> [accordion] What if an answer looks wrong?
>
> Tell your assistant what seems incorrect and ask it to record feedback. For help, contact the Quanta team.

> [!WARNING]
>
> AI-generated explanations can make mistakes. Check the reporting period, entity, accounting basis, and supporting figures before relying on an answer.

## Support

For help connecting or using the Quanta MCP server, contact the Quanta team at [support@usequanta.com](mailto:support@usequanta.com).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.