mcpField
E
ExpenseBot

ExpenseBot

Remote
expensebot.aiFinance0Updated 4w ago

About ExpenseBot

ExpenseBot provides expense management and receipt processing for AI assistants and automation platforms, including Gmail capture, expense organization, reports, and accounting handoffs. Its hosted remote MCP server exposes these tools, with write actions gated by previews, confirmations, or app handoffs.

Key features

  1. 01Submit receipts and check processing results
  2. 02Scan Gmail receipts and monitor progress
  3. 03Search and analyze expenses, income, and mileage
  4. 04Review, categorize, and correct expenses
  5. 05Create, export, and share reports
  6. 06Check compliance issues and account records

Use cases

  1. 01Collect receipts and handle duplicates or review holds
  2. 02Scan Gmail for new receipts
  3. 03Organize travel expenses and create a report
  4. 04Check client advance balances or expense trends

MCP support

The hosted remote MCP server exposes tools for expenses, receipts, reports, and related operations, with OAuth connection support. A local stdio adapter is also available for clients without remote OAuth support.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.expensebot.ai/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Packages
npm@expensebot/mcp-server-auth0.6.26

Tools

61
ParameterTypeRequiredDescription
tagsarrayNoFilter by tags
limitintegerNoResults per page (default 20, maximum 50)
querystringNoOptional natural-language recall over the user's existing expense fields (merchant, city/location, Notes including receipt items, delivery source, payer, and Business purpose, Tag, category). Deterministic case-insensitive matching — no embeddings or model classifiers. Combine with structured filters using AND semantics. Examples: 'Tribeca restaurant', 'Bodewell project hardware store', 'client dinner note'.
cursorstringNoOpaque nextCursor returned by the previous search_expenses page. Reuse the same filters; never construct or edit this value.
dateRangeanyNoTime period filter. Use exactly one variant — pick the shape that matches the user's phrasing.
maxAmountnumberNoMaximum expense amount
merchantsarrayNoFilter by merchant names (e.g., ['Uber', 'Starbucks'])
minAmountnumberNoMinimum expense amount
categoriesarrayNoFilter by expense categories (e.g., ['Travel', 'Meals'])
hasReceiptbooleanNoWhen true, return only expenses with a receipt link. When false, return only expenses without one.
clientEmailstringNoClient account email. Accountants may use this only for an accepted ExpenseBot client.

Details

From the project README

ExpenseBot logo

ExpenseBot MCP

Expense management and Gmail receipt capture for AI assistants and automation platforms.

Public installation and discovery metadata for the hosted ExpenseBot Model Context Protocol server.

ExpenseBot gives AI assistants and automation platforms structured access to expense data, Gmail receipt capture, reports, income, mileage, compliance checks, and accounting handoffs. ExpenseBot remains the system that captures and organizes receipts; downstream accounting software remains the source of truth for the books.

Receipt ingestion for agent workflows

ExpenseBot can also be the receipt-intake component inside a larger AI or automation workflow. Your agent owns orchestration, business rules, review policy, and downstream systems; ExpenseBot handles supported receipt intake, extraction, duplicate detection, terminal processing status, and customer-owned evidence in Google Sheets and Drive.

A reliable integration should:

  1. Submit each receipt once and retain the returned submission identifier.
  2. Check that identifier until it reaches an authoritative terminal result.
  3. Branch explicitly on success, duplicate, held-for-review, or error states.
  4. Avoid resubmitting work that is still pending.
  5. Connect each customer's own ExpenseBot account rather than sharing one personal credential across clients.

See the builder integration overview for the workflow contract, architecture boundary, and implementation examples.

Recommended: remote OAuth

Use the hosted Streamable HTTP endpoint in clients that support remote MCP and OAuth:

https://mcp.expensebot.ai/mcp

Example mcp.json:

{
  "mcpServers": {
    "expensebot": {
      "url": "https://mcp.expensebot.ai/mcp"
    }
  }
}

The client should open ExpenseBot sign-in and consent in the browser. Do not paste an ExpenseBot token into chat.

Local stdio fallback

Clients without remote OAuth support can run the public npm adapter:

npx -y @expensebot/mcp-server-auth

Store a revocable ExpenseBot token in the client's EXPENSEBOT_TOKEN environment or secret configuration. Generate and revoke tokens from AI Assistant Tokens.

What agents can do

  • Search and analyze expenses, income, mileage, subscriptions, and profit and loss.
  • Submit receipt images and PDFs and check processing results.
  • Scan connected Gmail accounts for receipts and monitor scan progress.
  • Add cash expenses, mileage, income, and confirmed client advances.
  • Review, correct, categorize, group, and annotate exact expenses with confirmation safeguards.
  • Create, inspect, export, and share reports and continue into client billing.
  • Review credits, refunds, compliance issues, monthly books, and client advance balances.
  • Trace document activity and open the relevant ExpenseBot review UI when a visual workflow is safer.

The live server currently exposes more than 50 tools. Write actions use previews, confirmations, bounded selections, or app handoffs where appropriate.

Useful prompts

  • Scan Gmail for new receipts and tell me what needs attention.
  • Show me my unsubmitted travel expenses from last month.
  • Group my Toronto Uber Eats expenses from this quarter under Client Demo, then create a report without personal expenses and help me bill the client.
  • How much of Client Acme's advance remains?
  • Compare this quarter's spending with the same quarter last year.

Documentation and support

This repository intentionally contains public installation metadata only. The hosted ExpenseBot application and its private source code are not published here.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.