Thread
The thing you carry from run to run: AI memory and context together. It keeps your project coherent across tools and models.
See also: Core, Working Set, Receipts.
The terms used in the method, in plain words.
The thing you carry from run to run: AI memory and context together. It keeps your project coherent across tools and models.
See also: Core, Working Set, Receipts.
A small file that always travels with the project: summary, standing rules, names/glossary. Kept small on purpose so it works with any model.
This is the stable part. It changes rarely.
Task-specific notes and docs you attach for the job at hand (design notes, research, retrieved pages). Drop it when you're done, the Core remains.
A short log for each change or decision: who · why · source · date (plus model/version). Lets you prove where things came from and pick up the thread later.
See also: Same Setup.
A repeatable arrangement: same Core, same Working Set, same instructions. Use it to compare models fairly or to re-run later.
Phrase it like: "Replay the same setup on another model."
The file that holds your Core. A portable AI memory document: summary, rules, names. When something sticks, it goes here.
In the templates: Core_Template.md
The running log where you record decisions and progress, each with receipts. Important things get promoted into the Brief over time.
In the templates: Addendum_Template.md
RAG (retrieval-augmented generation) pulls fresh facts into your Working Set. The Core and Receipts keep what should last. Use both together.
The Thread stays the same across ChatGPT, Claude, and whatever comes next. That's why the Core stays small and the receipts stay clear, no vendor lock-in.
Your files, your context, your control. The method is just markdown files that live where you put them. See the Privacy page for site-specific details.
A term for systems that capture your decision traces (the "why" behind actions) as platform infrastructure. The pitch: whoever owns your reasoning owns the relationship. Threadbaire's position is the opposite, your decision traces shouldn't become someone else's asset. The method keeps your receipts in files you control, portable across models, with no platform capturing your context.
Same insight, different owner: you.