# Curating Always-Loaded Context for LLM Agents: A Capacitated Assortment Model with Censored Feedback

arXiv: https://arxiv.org/abs/2610.11007 · HTML: https://mlsanity.com/paper/2610.11007/

## Key facts

- No citing papers in the map yet (data through 7 Oct 2026).
- Placed in Agent memory systems, one of 16 directions in LLM agents and tool use.
- Builds on 7 of its 42 references that are in the map.
- First posted on arXiv 7 Oct 2026.

## Placed in

- LLM agents and tool use / [Agent memory systems](https://mlsanity.com/agents/agent-memory-systems/)

## Builds on (most cited first, 7 of 7)

- [Evaluating AGENTS.md: Are Repository-Level Context Files Helpful for Coding Agents?](https://mlsanity.com/paper/2602.11988/) (2026)
- [How Many Instructions Can LLMs Follow at Once?](https://mlsanity.com/paper/2507.11538/) (2025)
- [Forget to Improve: On-Device LLM-Agent Continual Learning via Budget-Curated Memory](https://mlsanity.com/paper/2606.25115/) (2026)
- [Memory as a Controlled Process: Learned Adaptive Memory Management for LLM Agents](https://mlsanity.com/paper/2607.13591/) (2026)
- [Guardrails Beat Guidance: A Large-Scale Study of Rules, Skills, and Persistent Configuration for Coding Agents](https://mlsanity.com/paper/2604.11088/) (2026)
- [Instruction Adherence in Coding Agent Configuration Files: A Factorial Study of Four File-Structure Variables](https://mlsanity.com/paper/2605.10039/) (2026)
- [Why Does CLAUDE.md Keep Growing? Catastrophic Remembering in Agentic Coding](https://mlsanity.com/paper/2608.11095/) (2026)

## Abstract (arXiv)

At the start of every session, LLM agents load a fixed context file, such as textttAGENTS.md. Each loaded token in the file is charged again in every later round of the session, and these files can degrade performance as they grow in size. However, in practice, human or automated curators usually grow these files by appending. We formulate context curation as a capacitated assortment problem. Instructions consume tokens under a finite attention capacity; adding an instruction never raises the compliance of the others, while retained instructions incur a per-session setup cost. We prove an upper bound on the optimal file size, regardless of the number of available candidate instructions, and that appending every instruction with positive standalone value can be arbitrarily worse in net value than selecting an optimal subset. A token budget also limits the loss when the token price is underestimated. We then examine what can be learned from past sessions and how this information can guide decisions to add or remove instructions. Feedback is inherently censored: the benefits and harms of loaded instructions are observable, whereas missing instructions generate feedback only when their absence causes harm. In this setting, we show that deleting instructions ignored by agents can inevitably remove helpful ones. We characterize how much evidence should be collected before adding an instruction. Besides, we bound regret when human reviewers can inspect only a limited number of edits per period. Empirical experiments further show that irrelevant rules drawn from real context files reduce language-model compliance.