---
title: "Deep Research: Plain-Text Books vs Mercury"
created: 2026-10-12
visibility: public
status: live
type: research-summary
tags: [deep-research, bookkeeping, hledger, data-ownership]
summary: "hledger is still the best plain-text choice. Books by Mercury is free for 2026 and then $35 a month, with no full ledger export I could find, so it stays a second view and the plain-text books stay the record."
---

I had the bookkeeping question for my businesses researched properly. Is hledger still the best of the plain-text tools, what does Books by Mercury really do and cost, and how does each fit with owning my data.

hledger is still the one. It had about 2,000 commits and 17 releases in the last year, and nothing new displaces it for CSV rules, one repository per business and agents driving a command line. The newcomer worth watching is rustledger, which ships an MCP server, but it is nine months old and one person.

Mercury's price is $35 a month, not $30, after a free 2026. Nobody says it is capped, and it is not stated per business. Its API and agent layer are promised, not shipped, and I could not find a full ledger export. That matters for [[Data sovereignty for individuals]], because walking away and keeping the system is the whole point.

A weekly agent catch-up comes out at a few dollars a week across the businesses, close to my guess. The real cost is my attention for statements and exceptions. That is the same trade I made in [[How I chose my agent system in one day]], and it is why the choice already made for [[Main Branch]] holds.

So I keep hledger as the book for [[Awake Happy]], [[Noontide]] and the rest, and use Mercury's free year as a second view on one business, never as the truth. I look again in December. The point connects to [[Local-First Software]]: the system should still be mine if the vendor changes.

Full Deep Research:
[[Plain-Text Books vs Books by Mercury — Research Dossier]]