Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/xxyoudeadpunkxx/memory-assisted-shaping/llms.txt

Use this file to discover all available pages before exploring further.

Memory-Assisted Shaping is a lightweight behavioral protocol for ChatGPT Projects. It gives GPT a small continuity layer — tracking decisions, open gates, discarded paths, and source boundaries — so that long idea-shaping sessions don’t drift, and final artifacts stay clean and free of process residue.

Introduction

Learn what Memory-Assisted Shaping is, why it exists, and how it keeps sessions structurally coherent.

Quickstart

Set up the protocol in your ChatGPT Project and run your first shaped session in minutes.

How It Works

Understand the operating model, states, gates, and memory signals that power the protocol.

Python Tool

Reference for the notes.py append-only persistence helper and its full command set.

Why Memory-Assisted Shaping?

Long ChatGPT conversations drift. Examples become decisions. Discarded paths return. Open questions quietly disappear. By the time you ask for a final output, the artifact often carries traces of the entire process — drafts, reversals, exploratory tangents. Memory-Assisted Shaping gives GPT a structured way to track what matters without turning the conversation into file management. GPT follows real gates and source boundaries, uses append-only internal notes only when continuity would otherwise degrade, and produces a final artifact only after you explicitly approve it.

Operating States

Three explicit states gate every phase of a session — Reading, Shaping, and Synthesis.

Real Gates

A structured decision framework — ASK, PROPOSE, DEFER, STOP — ensures no gate is skipped or faked closed.

Memory Signals

Minimal append-only signals preserve continuity without making memory management the center of the conversation.

Authority Order

Operator instruction always dominates. A clear priority hierarchy prevents conflicts between context, protocol, and inference.

Lenses

Five simultaneous lenses — Idea, Shape, Implementation, Behavioral, Artifact — activate only when they change something meaningful.

Synthesis Discipline

Final artifacts are clean, standalone, and produced only after explicit approval — never through implied momentum or silence.

How to use these docs

1

Read the Introduction

Get oriented on the problem this protocol solves and how the operating model works end-to-end.
2

Follow the Quickstart

Add the protocol files to a ChatGPT Project and run a short session to see the gates and memory signals in action.
3

Study the concept pages

Deepen your understanding of operating states, real gates, and memory signals before customizing behavior.
4

Reference the protocol and Python tool

Use the Protocol Reference and Python Tool sections when you need authoritative detail on specific behaviors or CLI commands.

Build docs developers (and LLMs) love