LFM Methods

Methods

Methods

Things this laboratory knows how to build, written down so you can build them too. Each one is a guide you read and a kit your coding agent runs for you. Every revision is free once you have bought it.

The Memory Layer — Method and Kit

USD 49
  • Your coding agent forgets everything the moment you close it. This gives it a memory that lasts, and the judgement to keep only what is worth keeping.
  • Hand the kit to Claude Code, Claude Desktop, Codex or Cursor and it builds the whole thing for you
  • The code is one file of plain Python with nothing to install, and thirty-two tests so you can watch it work before you rely on it
  • A nine-page guide behind every decision, and every future revision free
Pay USD 49 and start

The Assistant — Method and Kit

USD 49
  • Something that has already done the work when you sit down in the morning: one job, done daily, without being asked.
  • Hand the kit to your coding agent and it builds the loop, the memory, and a service that starts itself again when your machine restarts
  • Anything it cannot take back waits for your approval first, which is the difference between an assistant and a liability
  • An eight-page guide, twenty-one tests, and every future revision free
Pay USD 49 and start

The Automation Audit — Method and Kit

USD 49
  • Find the work quietly eating your week, and prove which parts are worth automating before you pay for a single tool.
  • A sheet with the sums already in it: put in what you actually do, and it ranks what to fix first
  • Then your agent builds it safely, so running the same job twice never charges, emails or invoices anyone twice
  • A seven-page guide, seventeen tests, the note on which tools you may not resell, and every future revision free
Pay USD 49 and start

The Firemonkey Method — Method and Kit

USD 49
  • Stop getting work you have to fix. Ten habits that turn a coding agent from a vending machine into one that argues with you and hands back work already checked.
  • The kit asks what keeps going wrong, then writes the file your agent reads at the start of every session
  • The five parts of a brief that replace commands, with a worked example you can copy today
  • An eight-page guide, sixteen tests, and every future revision free
Pay USD 49 and start

All four Methods, with their Kits

USD 129
  • All four: a memory that lasts, an assistant built on it, the audit that says what to automate, and how to work with the agent doing it.
  • Four kits your agent runs, for Claude Code, Claude Desktop, Codex and Cursor
  • Every runnable file with its tests beside it, eighty-six in all
  • Thirty-two pages in all, normally USD 196 bought separately, and every future revision of everything free
Pay USD 129 and start

How it works

01

Pay

Every price is on the page. No call, no quote round, no invoice to chase.

02

Download it

Straight away, not within a day. The guide, the kit and the working code.

03

Your agent builds it

Point Claude Code, Claude Desktop, Codex or Cursor at the kit and it does the work.

If it goes wrong

If the work is not what the page promised, it is refunded. The catalogue stays short precisely so that stays true. Questions before or after buying go to hello@lfmlabs.com, and a person answers them.

We do it for you

Send a link and the finished files come back within a day. Same fixed prices, same no phone call, but ours are the hands on it.

We install the memory layer in your stack

USD 1,500
  • We build it into your codebase, in your language, wired to the agent you already use. One afternoon of ours instead of a fortnight of yours.
  • Search that finds things by meaning as well as by word, and a gate that keeps the noise out
  • Tests you own, and a written note on what we chose and why
  • Two weeks of questions afterwards, by email
Pay USD 1,500 and start