# BennyBooks

> AI-friendly accounting for small companies. Ledger primitives are transactions, notes, and receipts. Agents read `/api/v1` with a bearer token and write through an approval queue.

Use this file as the map. Fetch linked markdown only when you need the detail. Do not scrape HTML.

Auth: `Authorization: Bearer bb_live_<token>` from Settings → MCP access. Scopes: `read`, `propose-write`. Prefer `categoryCode` over category names. Never auto-post journal entries. Writes are audit-logged for humans in Log (`/log`) — no agent read API for Log in v1. For MCP bridges, enable only the curated tools (~12) — see [MCP tool design](https://bennybooks.com/developers/guides/mcp-design.md). List default limit is 50.

## Start here

- [Agent skill](https://bennybooks.com/skill.md): How to categorize, investigate, and close the books
- [Audit Log](https://bennybooks.com/developers/guides/audit-log.md): How writes appear in Log for human review
- [MCP tool design](https://bennybooks.com/developers/guides/mcp-design.md): Curated tools, annotations, lean responses
- [Product overview](https://bennybooks.com/index.md): What BennyBooks is and how agents should use it
- [Quickstart](https://bennybooks.com/developers/quickstart.md): First API calls
- [OpenAPI](https://bennybooks.com/openapi.json): Machine-readable REST contract with `x-mcp` (public)
- [Ontology JSON-LD](https://bennybooks.com/api/v1/ontology.jsonld): Category codes, counterparties, vocabularies (public)

## Docs

- [Introduction](https://bennybooks.com/developers.md): A structured REST API for AI agents and integrations. Read the ledger freely; write through an approval queue.
- [Quickstart](https://bennybooks.com/developers/quickstart.md): Make your first API call in under two minutes.
- [Authentication](https://bennybooks.com/developers/authentication.md): Bearer tokens, scopes, and security model.
- [Errors & responses](https://bennybooks.com/developers/errors.md): Response envelopes, error codes, and HTTP status mapping.
- [Connect with AnythingMCP](https://bennybooks.com/developers/guides/anythingmcp.md): Turn the curated BennyBooks OpenAPI surface into MCP tools for Cursor, Claude, and custom agents.
- [MCP tool design](https://bennybooks.com/developers/guides/mcp-design.md): How BennyBooks shapes MCP tools for clean, token-efficient agent use.
- [Connect as an AI agent](https://bennybooks.com/developers/guides/agents.md): Discovery files, markdown docs, OpenAPI, WebMCP, and how to call the ledger API.
- [Audit Log](https://bennybooks.com/developers/guides/audit-log.md): How agent writes and human approve/reject decisions are recorded for review in Log.
- [Categorize transactions](https://bennybooks.com/developers/guides/categorize.md): Recommended agent workflow using stable category codes, counterparties, and evidence.
- [Investigate a transaction](https://bennybooks.com/developers/guides/investigate.md): Gather full context before proposing changes.
- [Month-end checklist](https://bennybooks.com/developers/guides/month-end.md): Flag anomalies and workload items using dashboard summary.
- [Overview](https://bennybooks.com/developers/api.md): All v1 endpoints at a glance.
- [Transactions](https://bennybooks.com/developers/api/transactions.md): List, filter, and retrieve ledger transactions with related data.
- [Categories](https://bennybooks.com/developers/api/categories.md): Chart of accounts for categorization proposals.
- [Connectors](https://bennybooks.com/developers/api/connectors.md): Data source sync status and lineage.
- [Dashboard](https://bennybooks.com/developers/api/dashboard.md): Financial aggregates and agent workload indicators.
- [Proposals](https://bennybooks.com/developers/api/proposals.md): Approval-gated agent write operations.
- [Notes](https://bennybooks.com/developers/api/notes.md): Free-text context attached to transactions.
- [Receipts](https://bennybooks.com/developers/api/receipts.md): Search and retrieve receipt metadata.
- [OpenAPI spec](https://bennybooks.com/developers/reference/openapi.md): Machine-readable API contract with agent instructions and x-mcp annotations.
- [Data model](https://bennybooks.com/developers/reference/data-model.md): Core ledger types and relationships.
- [Ontology](https://bennybooks.com/developers/reference/ontology.md): Semantic model, stable category codes, counterparties, and agent categorization rules.

## Discovery

- [llms.txt](https://bennybooks.com/llms.txt): This file
- [Full docs corpus](https://bennybooks.com/llms-full.txt): Concatenated markdown for RAG
- [agents.txt](https://bennybooks.com/agents.txt): Capability announcement
- [agents.json](https://bennybooks.com/agents.json): Structured capability catalog
- [API catalog](https://bennybooks.com/.well-known/api-catalog): RFC 9727 linkset for the REST API
- [AI catalog](https://bennybooks.com/.well-known/ai-catalog.json): Links to docs, skill, OpenAPI, WebMCP pages

## Optional

- [HTML developer docs](https://bennybooks.com/developers): Same content as the markdown pages, for humans
- [AnythingMCP](https://bennybooks.com/developers/guides/anythingmcp.md): Turn curated OpenAPI into MCP tools without hosting a custom MCP server
- [MCP tool design](https://bennybooks.com/developers/guides/mcp-design.md): Which tools to enable and why
- [JSON-LD context](https://bennybooks.com/api/v1/ontology/context.jsonld): `@context` for expanding API resources