context-degradation

Detect context degradation patterns like lost-in-middle and poisoning in long-context samples.

11|1|Updated Jul 5, 2010

Search Tags:#degradation #lost-in-middle #poisoning #distraction

Ready to run

npx skills add https://github.com/georgeguimaraes/dotfiles --skill context-degradation-georgeguimaraes

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: context-degradation
Source: https://github.com/georgeguimaraes/dotfiles/tree/main/dot_claude/skills/context-engineering/context-degradation
Command: npx skills add https://github.com/georgeguimaraes/dotfiles --skill context-degradation-georgeguimaraes

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill helps identify degradation patterns that occur as context grows, enabling proactive mitigation of "lost in the middle," poisoning, distraction, and clash.

Core Concepts

Lost-in-Middle: attention bias toward edges
Context Poisoning: accumulation of incorrect info
Context Distraction / Clash: noise crowds out signal
Mitigations: compaction, masking, partitioning, isolation

Quick Start

Run the degradation detector on a long-context sample to identify lost-in-middle patterns.