accounting-mcp
Multi-tenant MCP server connecting accounting software to AI assistants via ~87 tools. Supports Bokio (Swedish accounting) with mock mode for development.
README
accounting-mcp
Multi-tenant remote MCP server that connects accounting software to MCP clients (Claude, ChatGPT, Perplexity, Claude Code). Business administrators sign in, connect their accounting company via OAuth, and their AI assistant can then work with their bookkeeping through ~87 tools.
Providers: Bokio (Swedish accounting SaaS). The core is provider-generic — adding Fortnox/Visma later means new files under src/providers + tool definitions, no changes to auth/transport/storage.
Architecture
pnpm/Turborepo monorepo with two deployable apps:
apps/api— the MCP + auth + billing server (Hono on@hono/node-server).apps/web— the user-facing dashboard (Next.js App Router). Talks to the api via tRPC (/trpc, router inapps/api/src/trpc/router.ts, type-only import in the web app) and proxies/api/*,/connect/*and/trpc/*to the api via Next rewrites, so the better-auth session cookie stays first-party on the web origin. The api'sWEB_URLenv points browser-facing redirects (sign-in, consent, Bokio OAuth callback) at the web origin. SeeDEPLOY.md.
The api is one stateful Node process with three faces:
- OAuth 2.1 authorization server toward MCP clients — better-auth +
@better-auth/oauth-provider: dynamic client registration, PKCE, sign-in/consent pages, token issuance. Metadata at/.well-known/oauth-authorization-server. - MCP resource server at
/mcp(Streamable HTTP via@hono/mcp) — validates our JWTs, one session perMcp-Session-Id, serves the tool set. - OAuth client toward Bokio —
/connect/bokioflow; company tokens AES-256-GCM-encrypted in Postgres, auto-refreshed with rotation + per-connection mutex.
Users authenticate with email/password; each user can connect multiple Bokio companies. Tools take an optional companyId (defaults when exactly one company is connected); tenant isolation is enforced in src/tools/company-resolver.ts.
Development
docker compose up -d # Postgres on localhost:5433
cp apps/api/.env.example apps/api/.env # fill BETTER_AUTH_SECRET + TOKEN_ENCRYPTION_KEY (openssl rand -base64 32)
pnpm install
pnpm --filter @acc/api db:migrate
pnpm --filter @acc/api dev # api on http://localhost:3000
pnpm --filter @acc/web dev # dashboard on http://localhost:3001 (optional; api serves MCP alone)
With BOKIO_MOCK=true (the default in .env.example) a mock Bokio — OAuth endpoints + fixture companies with invoices, journal entries, chart of accounts, SIE export — is mounted at /mock/bokio/v1. The entire product works end-to-end with zero Bokio credentials.
Verify
pnpm turbo run build typecheck test # all workspaces
pnpm --filter @acc/api e2e # against a running api: token connect suite + full
# DCR → OAuth → consent → MCP session → tools → SIE flow
Connect from Claude Code: claude mcp add --transport http accounting http://localhost:3000/mcp (the OAuth flow opens in your browser — sign up, approve, then connect a mock company from /dashboard).
Configuration
See .env.example. Highlights:
BOKIO_ALLOW_WRITES(defaultfalse) — when off, mutating tools are not even listed. Write tools carry MCPdestructiveHint/readOnlyHintannotations either way.BOKIO_MOCK— mock Bokio instead ofapi.bokio.se.TOKEN_ENCRYPTION_KEY— base64 32-byte AES key for provider tokens at rest.BOKIO_SCOPES— defaults to every scope documented in the spec (invoices:read,journal-entries:write, …).
Going live with real Bokio
- Create a developer account + integration at the Bokio developer portal; register redirect URI
https://<host>/connect/bokio/callback. - Set
BOKIO_CLIENT_ID/BOKIO_CLIENT_SECRET,BOKIO_MOCK=false,BASE_URL=https://<host>. - For local testing against real Bokio, tunnel with
cloudflared tunnel --url http://localhost:3000and use the tunnel URL asBASE_URL.
Deployment (Railway or any container host)
Multi-stage Dockerfile; the image is self-contained. On Railway: add a Postgres service, set the env vars, use npm run db:migrate as release command, health check GET /healthz. Note: MCP sessions are in-memory — run a single instance (or add sticky sessions before scaling out).
Layout
src/
auth/ better-auth config (OAuth AS) + Bokio connect flow
http/ Hono app, /mcp session handling, HTML pages (sign-in/consent/dashboard)
bokio/ generated OpenAPI types, API client (401-retry), token-manager, mock server
tools/ declarative registry, meta tools, per-resource Bokio tool tables
db/ Drizzle schema (better-auth tables + accounting_connection + tool_audit_log)
lib/ AES-256-GCM token crypto
specs/ Bokio OpenAPI specs (source for npm run gen:bokio)
scripts/ e2e-oauth.mjs — full-flow integration test
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.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
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.