---
source: https://iinventedlight.com/library/notes/ao-context-evolving-playbooks.md
title: "Grow agent context as a playbook, not a rewritten summary"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
source_license: "CC BY 4.0; restated in our own words"
attribution: "Summary by iinventedlight (AI-assisted). Sources: arXiv, [2510.04618] Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models, CC BY 4.0; arXiv, [2505.06120] LLMs Get Lost In Multi-Turn Conversation, CC BY 4.0."
author: "iinventedlight (synthesis of cited sources); text drafted with AI assistance"
date: 2026-11-04
status: published
checked_on: 2026-11-04
as_of: 2026-11-04
confidence: medium
untested: "No experiment was rerun; only the fetched arXiv abstract pages were read. Results are as reported by the authors."
category: agent-operations
topic_area: "agent operations: context"
tags: agents, context-engineering, playbooks, memory
sources:
  - id: S1
    title: "[2510.04618] Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models"
    publisher: arXiv
    url: https://arxiv.org/abs/2510.04618
    accessed: 2026-11-04
    license: "CC BY 4.0"
  - id: S2
    title: "[2505.06120] LLMs Get Lost In Multi-Turn Conversation"
    publisher: arXiv
    url: https://arxiv.org/abs/2505.06120
    accessed: 2026-11-04
    license: "CC BY 4.0"
---
# Grow agent context as a playbook, not a rewritten summary

## Short answer
Update prompts and memory with small structured additions instead of repeated rewrites. Rewrites cause brevity bias and context collapse. ACE's incremental playbooks beat baselines by 10.6% on agent tasks. [S1]

## Caution
Long accumulating chats can hurt too: models often lock onto early guesses. [S2] Curate what is kept; do not just append.

## Claims
- id: C1
  claim: Iterative rewriting erodes details.
  source: S1
  evidence: "context collapse, where iterative rewriting erodes details over time"
- id: C2
  claim: ACE uses structured, incremental updates.
  source: S1
  evidence: "ACE prevents collapse with structured, incremental updates"
- id: C3
  claim: ACE improved agent results by 10.6%.
  source: S1
  evidence: "+10.6% on agents"
- id: C4
  claim: LLMs make early assumptions and over-rely on them.
  source: S2
  evidence: "LLMs often make assumptions in early turns"

## Sources
- [S1] [2510.04618] Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models. arXiv. https://arxiv.org/abs/2510.04618 . License: CC BY 4.0. Accessed 2026-11-04.
- [S2] [2505.06120] LLMs Get Lost In Multi-Turn Conversation. arXiv. https://arxiv.org/abs/2505.06120 . License: CC BY 4.0. Accessed 2026-11-04.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted).