# Context Rot

> The gradual loss of instruction-following quality as a conversation or agent trace fills with stale, noisy, or contradictory context.

- Category: Context & Memory
- Stage: Emerging
- Also called: Context degradation, Long-context drift
- Canonical: https://www.joelcsaji.com/ai-dictionary/context-rot

## Plain English

Too much leftover chat junk makes the model forget what actually matters.

## Example

After dozens of tool results, an agent ignores the original acceptance criteria buried near the top of the thread.

## Why it matters

Long-running agents need compaction, resets, and memory design, not just a bigger context window.

## Related

- [Compaction](https://www.joelcsaji.com/ai-dictionary/compaction)
- [Context Truncation](https://www.joelcsaji.com/ai-dictionary/context-truncation)
- [Context Engineering](https://www.joelcsaji.com/ai-dictionary/context-engineering)

## Sources

- [Anthropic: Effective Harnesses](https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents)
- [OpenAI: Agents SDK](https://platform.openai.com/docs/guides/agents-sdk)