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.

This guide walks you through placing the three protocol files into a ChatGPT Project, initializing a session with the Python helper, activating the protocol in your conversation, and shaping your first idea without losing decisions, gates, or discarded paths along the way.
1

Download the files

Get protocol.md, note-io.md, and notes.py from the repository. Keep them together in a single directory. The recommended layout is:
memory-assisted-shaping/
  README.md
  protocol.md
  note-io.md
  notes.py
Generated session files (session_meta.json, session_notes.jsonl, session_export.json) will be written into this same directory during use. Do not commit generated session files as part of the clean source package.
2

Add files to ChatGPT Project Source

In ChatGPT Projects, add protocol.md, note-io.md, and notes.py as project source files. GPT reads all three at session start — the behavioral contract, the I/O boundary contract, and the Python helper — so it knows the protocol, the signal format, and where persistence lives before you say anything.
3

Initialize the session (optional Python)

From the directory containing notes.py, run:
python3 -S ./notes.py init
This creates session_meta.json with a UUID session ID and an empty session_notes.jsonl ready to receive signals. This step is optional — if you skip it, the first append call initializes the session automatically.
4

Activate the protocol in your conversation

Start your ChatGPT session and send this exact prompt:
Read protocol.md and use GPT-Memory-Assisted Shaping for this session.
GPT reads the protocol, enters READING_ALIGNMENT state, locates notes.py, and is ready to shape your idea. It will not repeat the session entry procedure on every turn.
5

Shape your idea

Work normally. Describe your idea, respond to GPT’s questions, make decisions, and rule things out. GPT emits memory signals only when continuity is at risk — a real decision is confirmed, a gate opens or closes, a path is discarded, or the shape moves in a material direction. You will not see memory management interrupt the conversation; GPT hides receipts unless they are useful or you ask.A typical append command that GPT would run internally looks like this:
python3 -S ./notes.py append --type decision --text "The final artifact is separate from shaping notes."
When an effect needs to carry forward into later shaping or synthesis, GPT adds --effect:
python3 -S ./notes.py append --type gate --text "Persistence model is deferred." --effect "Do not finalize the artifact until this is resolved."
6

Ask for a recap or final artifact

At any point you can ask GPT to recover the current session state. Useful recap prompts from the README:
  • “Recap the internal notes.”
  • “Show me the current shape of the idea.”
  • “List decisions, open points, discarded paths, and artifact readiness.”
  • “Are there any gates still open before we write the final artifact?”
When you are satisfied with the shape, give explicit approval and ask for the final artifact. GPT will produce a clean output — no process notes, no raw memory signals, no shaping rationale — from the consolidated shape.
Protocol-only mode: If Python execution is unavailable in your environment, GPT states once that append-only persistence is not active and then continues shaping normally under the protocol. Decisions, gates, and discarded paths are still tracked structurally in the conversation — they just will not be persisted to a JSONL log.
Before requesting the final artifact, run summary to verify the current session state and confirm that no gates are still open:
python3 -S ./notes.py summary
The summary returns the session ID, the total number of appended signals, a count broken down by signal type, and the most recent note. Reviewing this before synthesis helps you catch any unresolved gates or deferred decisions before they affect the artifact.

Build docs developers (and LLMs) love

Get started for freeTalk to us