agentleFS
median-labsmedian-labs/close-the-books/skills/stop-the-automation/SKILL.md

stop-the-automation

Skill1 starsChanged 2 months agoMITView on GitHub

Turn off the bank rules that are still posting into the year being filed, and find the entries they already posted with nobody watching. Use before any figure is produced for a return, and whenever a file has been left alone for months.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not describe a flagged entry as fraud or as an error.

    line 77

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Stop the automation

A company stopped being worked in April. Its bookkeeper left, its books went quiet, and every person who had ever touched the file moved on. In December a 3,187.65 deduction appeared in the tax year, on an expense account, in a month holding no other entry, with nothing behind it. A bank rule had matched a descriptor and posted it, eight months after anyone was watching.

That entry is inside the year on the return. Nothing about the queue, the balances or the trial balance would raise it: it balances, it foots, and it sits in an account where entries like it belong.

Hard gates

  • Nothing that produces figures for the return runs while rules are still posting into it. entries and handoff refuse. Every hour worked before the rules stop is worked against a file that is still changing under it.
  • Export the rules before turning them off. Rules, then Export rules. The export is how they go back afterwards, and it is also the only record of what was posting.
  • Never state who posted an entry unless an audit log says so. An entry alone in a quiet month that lands exactly where a rule would put it was either posted by automation or entered by one person in a month nobody else touched, and no export separates the two. Say what was measured and name the audit history as what settles it.
  • A flagged entry is a question, not a reversal. It gets removed or supported by the owner, on evidence, like any other entry.

Run it

python3 bin/books.py intake --last-human 2026-02-28 --note "the bookkeeper left"
python3 bin/books.py intake --rules 15 --auto-add 15

With auto-add above zero, entries and handoff refuse and name the count. The refusal also lists what has already posted into the filing year with nothing to say a person made it, largest first.

In QuickBooks: the gear, Rules. Turn off "Automatically confirm transactions this rule applies to" on each rule, or delete them. Then:

python3 bin/books.py intake --rules 15 --auto-add 0

A rule that only categorizes is not the same as a rule that posts. Rules can stay; auto-add cannot, while the year being filed is still open.

The four signals behind a flagged entry

  1. Alone in its period. The only entry its month holds. A month somebody worked has a spread of entries in it.
  2. No attachment where the account's other entries have them. Needs attachment data, which no QuickBooks xlsx export carries. Supplied or skipped, never inferred.
  3. An exact pattern match. The counterparty has landed in that account often enough that a rule could be written from the company's own history, and this entry lands there too.
  4. A poster naming itself. The strings an integration or a bank rule writes into a memo that a person would not type.

An audit log replaces all four. Pass one and the author is stated as a fact rather than inferred.

What to tell the owner

Two sentences. What is still posting, and what it has already put in the year they are about to file. Then the one thing to do: turn off auto-add, export the rules first.

Do not describe a flagged entry as fraud or as an error. It is an entry with nothing behind it, dated inside the year on the return, and the fix is either a document or a reversal.

What good looks like

Auto-add at zero on every rule, recorded with the date. The rules file exported and in the working directory. Every entry flagged in the filing year either supported by a document or reversed, with the decision recorded.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.