AgenticBooks MCP Server
Financial data infrastructure for AI agents. Connect to a startup's books to read live P&L and bank balances, review and reclassify transactions, manage the chart of accounts, and connect banking sources.
README
AgenticBooks MCP Server
Financial data infrastructure for AI agents. Connect Claude (or any MCP client) directly to a startup's books — read live P&L and bank balances, review and reclassify transactions, manage the chart of accounts, and connect banking sources — over a secure, OAuth-authenticated remote MCP server.
This is a hosted (remote) MCP server. You don't install or self-host it — you connect your MCP client to the AgenticBooks endpoint and authenticate with your AgenticBooks account. This repository is the public listing and connection guide; the server itself runs as managed infrastructure.
- Website: https://agenticbooks.com
- MCP endpoint:
https://mcp.agenticbooks.ai/mcp - Transport: Streamable HTTP
- Auth: OAuth 2.0 (Claude "Custom Connector" flow via Clerk) or an
ab_API key
What it does
AgenticBooks ingests webhooks from Stripe, Mercury, RevenueCat and other sources, normalises and classifies the events into double-entry accounting, and syncs to QuickBooks/Xero. This MCP server exposes that ledger to an AI agent so the agent can answer financial questions and take bookkeeping actions on the user's behalf — every write is audit-trailed and scoped to the caller's organisation.
Typical prompts once connected:
- "What's our P&L for last month?"
- "How much cash do we have across all bank accounts right now?"
- "Show me transactions that still need review, grouped by vendor."
- "Reclassify the Vercel charges as Hosting, and remember that for next time."
Connect
Claude (Desktop / Web) — Custom Connector (OAuth)
- Open Settings → Connectors → Add custom connector.
- Enter the MCP server URL:
https://mcp.agenticbooks.ai/mcp - Complete the OAuth sign-in when prompted. You'll authenticate with your AgenticBooks account; access is scoped to your organisation.
See Anthropic's guide to remote custom connectors: https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp
API key (any MCP client over HTTP)
If your client doesn't support the OAuth connector flow, authenticate with an
ab_ API key (create one in the AgenticBooks dashboard) as a bearer token:
{
"mcpServers": {
"agenticbooks": {
"transport": "http",
"url": "https://mcp.agenticbooks.ai/mcp",
"headers": {
"Authorization": "Bearer ab_your_api_key_here"
}
}
}
}
The two auth modes are independent — OAuth and ab_ keys both work; you only
need one.
Tools
All tools operate within the authenticated organisation. Reads are non-destructive; writes are recorded in an append-only audit trail.
Reporting & balances
| Tool | What it does |
|---|---|
get_financial_summary |
One-call snapshot: current-month P&L, balances across all connected providers, unreviewed-event count, and ledger sync health. |
get_pnl_report |
Profit & loss report over a date range, built from classified events. |
get_account_balance |
Current balances across the org's connected money providers. |
get_bank_balances |
Bank holdings, one line per connected bank (Mercury, etc.). |
get_audit_log |
Reads the append-only audit trail, newest first. |
Review & classification
| Tool | What it does |
|---|---|
get_unreviewed_events |
Lists classified events the rules engine couldn't confidently categorise. |
get_pending_by_counterparty |
Summarises the review queue clustered by counterparty and direction. |
approve_classification |
Approves one event from the review queue by assigning its account. |
reclassify_entry |
Changes the account on an already-posted ledger entry (audit-trailed). |
Counterparty rules (the learned vendor map)
| Tool | What it does |
|---|---|
create_counterparty_rule |
Creates/updates a persistent rule mapping a counterparty to an account. |
list_counterparty_rules |
Lists the org's learned counterparty → account rules. |
update_counterparty_rule |
Re-points an existing rule at a different account (affects future events). |
delete_counterparty_rule |
Removes a rule; future events from that counterparty stop auto-classifying. |
Chart of accounts
| Tool | What it does |
|---|---|
list_chart_accounts |
Lists the org's chart of accounts (code, name, type, active). |
add_chart_account |
Adds a new account; the system auto-assigns the code. |
rename_chart_account |
Renames an account (the code never changes). |
set_chart_account_active |
Deactivates or reactivates a chart account. |
Periods & integrations
| Tool | What it does |
|---|---|
close_period |
Runs the month-end close, including unrealized FX revaluation. |
reimport_historical |
Queues a historical import over a chosen window. |
connect_mercury |
Connects the org to Mercury via a read-only, scoped API token. |
disconnect_mercury |
Disables the org's Mercury integration (audit-trailed). |
Security
- Every tool call is authenticated (OAuth access token or
ab_API key) and scoped to a single organisation. - All writes are recorded in an append-only audit trail.
- Banking credentials are stored per-organisation and encrypted at rest; they are never accepted as, or exposed via, MCP tool arguments.
Links
- Website — https://agenticbooks.com
- Support — https://agenticbooks.com (contact via dashboard)
AgenticBooks is a hosted service. This repository documents how to connect; it is not the server source.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.