My agent consistently ignores instructions I place in the middle of long system prompts. It follows the beginning and end perfectly but misses middle sections. How do I diagnose and fix this? - [ ] Identifies "lost-in-middle" degradation pattern - [ ] Explains U-shaped attention curve (beginning/end favored) - [ ] Recommends restructuring critical info to beginning/end positions - [ ] Suggests context partitioning or chunking - [ ] May reference context-degradation.md for detailed patterns
Scanned 9/20/2026
Install to Claude Code
npx -y skills add tstapler/dotfiles --skill debug-lost-in-middle --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Debug Lost In Middle?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/tstapler-debug-lost-in-middle)More formats (shields.io, HTML) on the badges page.
---
name: debug-lost-in-middle
description: ''
---
# Prompt
My agent consistently ignores instructions I place in the middle of long system prompts. It follows the beginning and end perfectly but misses middle sections. How do I diagnose and fix this?
# Expected
- [ ] Identifies "lost-in-middle" degradation pattern
- [ ] Explains U-shaped attention curve (beginning/end favored)
- [ ] Recommends restructuring critical info to beginning/end positions
- [ ] Suggests context partitioning or chunking
- [ ] May reference context-degradation.md for detailed patternsIs this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!