Add failure pattern 11: [your pattern here]

[YuhaoLin2005]

What

Contribute a new AI agent failure pattern to docs/failure-patterns.md.

The current 10 patterns catalog AI agent failure modes observed in real Claude Code sessions. Each pattern follows this format:

### Pattern N: [Name]
- **What**: [One-line description]
- **Example**: [Real session excerpt, anonymized]
- **Why it happens**: [Root cause]
- **How to detect**: [Method — mechanical or behavioral]
- **How to fix**: [Actionable fix]

Ideas for Pattern 11

• Context blindness: agent ignores CLAUDE.md after conversation gets long
• Tool overreach: agent uses complex tool chain when simple command would work
• Silent assumption: agent assumes file state without checking
• Loop fixation: agent repeats same failed approach without escalation
• Your own pattern: something you've observed in your sessions

Notes

• See existing patterns
• Open an issue first to discuss before PR
• Read CONTRIBUTING.md