---
schema: 1
type: term
title: '**context rot**'
description: >-
  **Context rot** is the degradation that occurs when an AI system's working
  context accumulates too much material — instructions, intermediate outputs,
  and handoff details pile up until the model's output quality drifts. The term
  is used for the problem that architectures like _multi-agent handoff_ are
  designed to avoid.
canonical: 'https://liquidgold.cc/channel/everyinc/explore/term/context-rot-323c92256945'
markdown: >-
  https://liquidgold.cc/channel/everyinc/explore/term/context-rot-323c92256945.md
llms: null
parent: 'https://liquidgold.cc/channel/everyinc/llms.txt'
channel: everyinc
updated: '2026-10-23T22:12:35.346Z'
language: null
children: []
---


# \*\*context rot\*\*

Canonical: https://liquidgold.cc/channel/everyinc/explore/term/context-rot-323c92256945

Parent: [Channel index](https://liquidgold.cc/channel/everyinc/llms.txt)

## Definition

**Context rot** is the degradation that occurs when an AI system's working context accumulates too much material — instructions, intermediate outputs, and handoff details pile up until the model's output quality drifts. The term is used for the problem that architectures like _multi-agent handoff_ are designed to avoid.

**Context rot** describes what happens when the working context an AI system relies on degrades as it grows — instructions and intermediate outputs accumulate, and the quality of what the model produces drifts. In this channel's material, the term appears when discussing how to *engineer around* that degradation rather than just pushing more material into one long context.