# What Context Does a Coding Agent Actually Need to Act?

arXiv: https://arxiv.org/abs/2607.09691 · HTML: https://mlsanity.com/paper/2607.09691/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use.
- Builds on 11 of its 12 references that are in the map.
- First posted on arXiv 19 Jun 2026.

## Placed in

- LLM agents and tool use / [Software engineering coding agents](https://mlsanity.com/agents/software-engineering-coding-agents/)

## Builds on (most cited first, 11 of 11)

- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [From Local to Global: A Graph RAG Approach to Query-Focused Summarization](https://mlsanity.com/paper/2404.16130/) (2024)
- [RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval](https://mlsanity.com/paper/2401.18059/) (2024)
- [RepoGraph: Enhancing AI Software Engineering with Repository-level Code Graph](https://mlsanity.com/paper/2410.14684/) (2024)
- [LocAgent: Graph-Guided LLM Agents for Code Localization](https://mlsanity.com/paper/2503.09089/) (2025)
- [Sufficient Context: A New Lens on Retrieval Augmented Generation Systems](https://mlsanity.com/paper/2411.06037/) (2024)
- [Alibaba LingmaAgent: Improving Automated Issue Resolution via Comprehensive Repository Exploration](https://mlsanity.com/paper/2406.01422/) (2024)
- [SWE-Pruner: Self-Adaptive Context Pruning for Coding Agents](https://mlsanity.com/paper/2601.16746/) (2026)
- [ContextBench: A Benchmark for Context Retrieval in Coding Agents](https://mlsanity.com/paper/2602.05892/) (2026)
- [Compressing Code Context for LLM-based Issue Resolution](https://mlsanity.com/paper/2603.28119/) (2026)
- [CodeCompass: Navigating the Navigation Paradox in Agentic Code Intelligence](https://mlsanity.com/paper/2602.20048/) (2026)

## Built on by (most cited first, 3 of 3)

- [CodeNib: A Multi-View Data System for Serving Repository Context to Coding Agents](https://mlsanity.com/paper/2607.25431/) (2026)
- [The Recall Trap: A Recall-Maximizing Retriever Configuration Reduces Issue Resolution in Fixed-Budget Code Context](https://mlsanity.com/paper/2608.14838/) (2026)
- [Cortex: Measurement-Gated Evolution and Pruning of Coding-Agent Skills and Rules](https://mlsanity.com/paper/2610.14216/) (2026)

## Abstract (arXiv)

A modern coding agent can hold an entire repository in its context window. Most of its reading is wasted -- and the interesting question is not how much context an agent can use, but what it actually needs. We study that question at the moment it matters most: when the agent must edit code. Separating finding the work site from acting on it, we hold localization fixed with an oracle, vary only how the code is represented, and score context against real issue resolution on SWE-bench Verified. The answer is starkly minimal. The signal lives in the code being edited itself: natural-language summaries of it answer almost none of the behavioral questions that the source answers (4/45 vs. 27/45, held-out repositories, independent judge), and the gap belongs to the representation, not the summarizer -- a frontier model's summaries score exactly as poorly as a 3B model's. The surrounding context hardly matters either: across every multi-file instance in Verified, under a protocol frozen before any data, rendering a file's remainder as UML skeletons and signatures resolves no more issues than deleting that remainder outright (N=70, exact McNemar p=0.75). That was our registered hypothesis, and it failed. Compressed context, meanwhile, matches whole files at a third of the tokens: a resolved issue costs 19K context tokens, not 94K. The instrument also yielded a finding the field should keep: temperature-0 API inference flips ~9% of per-instance outcomes between byte-identical runs. That is a noise floor under every small effect reported on this benchmark, including ours. We release the instrument -- gold-validated environments, per-instance proof that every reference edit is expressible from every arm's context, deterministic patch construction, and pre-registered hypotheses whose nulls we publish.