---
title: "GPT-5 on context-budget rituals"
url: https://memory.wiki/EyW4HU4-
updated: 2026-05-14T18:15:49.480Z
hub: https://memory.wiki/hub/demo
concept_count: 10
source: "memory.wiki"
---
# GPT-5 on context-budget rituals

> ChatGPT-5 share link, 2026-04-09. I'd been losing context across sessions and wanted a framework, not another "use Projects" answer.

## The prompt

I dumped four things: the v6 AGENTS.md draft, the `mdfy Foundations` bundle, the most recent 50 Cursor chat turns, and a one-paragraph statement of what I was trying to design (cross-AI context handoff). Then asked: "How would you structure the context you carry into a working session?"

## What it built

A worksheet, not a heuristic. The structure was:

| Slot | Allocation | Source |
|---|---|---|
| Project context | 30% | AGENTS.md / CLAUDE.md / hub URL |
| Recent edits | 30% | Last N file diffs |
| Live conversation | 40% | This turn + last 4-6 |

The split is the part I actually use. Before this, I was loading "everything" — which meant the model got 12K tokens of static context and 1K of live discussion. The flip (live > context) sounds backwards but explains why "the AI keeps losing the thread mid-conversation" — the thread had no room.

## The caveat

The worksheet over-fits to **coding** sessions. For research / writing, the ratios invert: live conversation shrinks (you're not iterating on a turn-by-turn loop) and project context expands (you're loading more reference material).

## What I'm doing

- Default `?compact` on hub URLs when pasted into Cursor — claws back 30% of the project-context slot for free.
- Added a "Recent edits" Cmd+K command in the editor — pulls last 5 diffs into a paste-able block.
- For research, swap to `?full=1` on bundle URLs — the whole bundle as context, no compaction.


---

## Concepts in this document
- **Cursor** _(entity)_
  Code editor that consumes mdfy bundles as context for chat and composer sessions.
- **Token budget** _(concept)_
  The finite context window constraint that makes allocation strategy necessary and directly impacts model performance.
- **Context-budget allocation** _(concept)_
  Core framework dividing limited token space between static project data, recent changes, and live conversation.
- **Live conversation ratio** _(concept)_
  The 40% allocation to recent turns that counterintuitively dominates over static context in coding workflows.
- **Coding sessions** _(tag)_
  The primary use case where the 40% live conversation ratio proves effective.
- **Project context slot** _(concept)_
  The 30% allocation for static reference material like documentation and codebase summaries.
- **GPT-5** _(entity)_
  The AI model used to generate the context-structuring worksheet being evaluated.
- **Research and writing workflows** _(tag)_
  Alternative use case requiring inverted ratios with expanded project context and reduced live conversation.
- **Cross-AI context handoff** _(concept)_
  The design goal of maintaining coherent context when working across multiple AI systems in sequence.
- **Compact URL mode** _(concept)_
  Token-saving technique that reclaims 30% of project-context allocation through stripped hub URLs.

## Concept relations (within this doc's concepts)
- **Token budget** constrains **Context-budget allocation**
- **Research and writing workflows** requires **Project context slot**
- **Context-budget allocation** defines **Live conversation ratio**
- **Context-budget allocation** defines **Project context slot**
- **GPT-5** generated **Context-budget allocation**
- **Cursor** implements **Context-budget allocation**
- **Coding sessions** requires **Live conversation ratio**
- **Compact URL mode** optimizes **Project context slot**

_Hub canonical:_ https://memory.wiki/hub/demo
_Concept digest:_ https://memory.wiki/raw/hub/demo?digest=1&compact=1