What Context Does a Coding Agent Actually Need to Act?

• URL: http://arxiv.org/abs/2607.09691v1
• Date: Fri, 19 Jun 2026 04:22:17 GMT
• Title: What Context Does a Coding Agent Actually Need to Act?
• Abstract summary: A modern coding agent can hold an entire repository in its context window.<n>We study when the agent must emphedit code.<n>We score context against real issue resolution on SWE-bench Verified.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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 \emph{needs}. We study that question at the moment it matters most: when the agent must \emph{edit} code. Separating \emph{finding} the work site from \emph{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 $19$K context tokens, not $94$K. The instrument also yielded a finding the field should keep: temperature-0 API inference flips ${\sim}9\%$ 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.

Related papers

• Don't Let the Model Write the YAML: Deterministic, Minimal-Diff GitOps Remediation from LLM-Proposed Field Changes [0.0]
  In GitOps, having the model author the edited file or a diff is what practitioners reach for first.<n>We present an alternative that separates the semantic decision (which resource, field, and value) from the syntactic act of editing the file.<n>Because the file is never re-serialized, the diff is minimal by construction, and the edit is correct and deterministic independent of the model.
  arXiv  Detail & Related papers  (2026-08-31T18:37:12Z)
• Facts Without Rules: Boundary Metadata Collapse in Multi-Agent LLM Handoffs [13.92589238677081]
  We show that summaries preferentially preserve operational facts while weakening the boundary metadata that governs how those facts may be used.<n> Boundary-marker and operational-fact survival are nearly uncorrelated at the handoff level on both GPT-5-mini and DeepSeek-R1-32B.
  arXiv  Detail & Related papers  (2026-08-29T03:38:35Z)
• Long-Horizon State Tracking in LLMs: Executing MD5 through a Deep Sequence of Dependent Tool Calls [0.5097809301149342]
  Long-horizon tasks remain uncommon in large language model (LLM) evaluation.<n>We test whether an LLM can carry exact intermediate state across many tool calls at all.<n>Two ingredients decide success: keeping the model's own reasoning in its context each turn, and voting over a thinking-enabled worker to remove its modular-arithmetic slips.
  arXiv  Detail & Related papers  (2026-08-02T09:38:02Z)
• Teach it to stop, not just to click [0.0]
  We show a repaired policy's success rate is dominated by upstream footing.<n>We release a library (cua_reliability) for routine k-seed reporting.
  arXiv  Detail & Related papers  (2026-07-19T08:46:06Z)
• Calibration, Not Compilation: Detecting and Repairing Misspecified Probabilistic Programs Written by Language Models [31.082191748525137]
  Language models increasingly write probabilistic programs (in NumPyro, Stan, or Pyro)<n>A program that compiles, runs, and passes every unit test can still be emphstatistically wrong.<n>The right verifier is therefore not a test suite but the Bayesian workflow itself.
  arXiv  Detail & Related papers  (2026-06-30T13:16:39Z)
• Does the Same Token Mean the Same State? MoE Routing as Signal for Reasoning Control [53.43997385459622]
  We operationalize textbfRAD (Routing Agreement Decoding) as an answer-string-free multi-rollout selector.<n>It locates a fixed anchor, represents each rollout by its anchor-window MoE routing states, and returns the densest Weighted-Jaccard $K$-NN route-basin center.
  arXiv  Detail & Related papers  (2026-06-22T03:17:31Z)
• Isolating LLM Alignment from Regex: Zero Coverage and Metric-Dependent Divergence Under Adversarial Mutation [51.56484100374058]
  Previous work found no measurable coverage gain from adding a live Gemini backend behind an active filter.<n>We introduce $L_5$-no-regex -- identical to $L_4$-real (Gemini-2.5-flash, token-budget cap, rate limit, output scrub) but with the nine-pattern filter disabled.<n>We evaluate it against adversarial probes across three sub-languages.
  arXiv  Detail & Related papers  (2026-06-12T18:54:24Z)
• GitOfThoughts: Version-Controlled Reasoning and Agent Memory You Can Replay, Diff, and Merge [0.02066679446411895]
  We introduce GitOfThoughts, which stores an agent's reasoning tree as a git repository.<n>Every scored thought is a commit, scores are notes, outcomes are tags, and retrieval is "git log" over the agent's own history.<n>This makes reasoning replayable, auditable, and mergeable across agents at near-zero engineering cost.
  arXiv  Detail & Related papers  (2026-06-12T14:02:37Z)
• Selection Integrity for LLM Graph Memory: An Accumulability Criterion for Information-Flow-Blind Retrieval [16.858503871105135]
  We show that the entire class of provenance defense is blind by construction.<n>A long-term graph memory runs a global selection step over writable graph structure.<n>We prove the immune case in general and the open case under a chokepoint condition we verify.
  arXiv  Detail & Related papers  (2026-06-10T16:26:46Z)
• Enhancing Software Engineering Through Closed-Loop Memory Optimization [50.418699819003486]
  We introduce ours, a closed-loop framework for memory augmentation in software engineering (SE) agents.<n>ours grounds memory utility in textitvalidated downstream impact, establishing utility as a task-agnostic textbf benchmark and an annotation-free textbfoptimization signal.<n>Results demonstrate that ours consistently improves SE agents across settings, achieving absolute gains of up to $uparrow5.25%$ in success rate and $uparrow4.63%$ in resolve efficiency.
  arXiv  Detail & Related papers  (2026-06-04T03:17:21Z)
• Subtle Injection for Ground-truth Inference of LLM Training Data [0.2538209532048867]
  SIGIL is a framework that embeds imperceptible emphcanary sequences into protected text and code.<n> SIGIL defines five canary strategies -- lexical-rare, lexical-phrase, syntactic, semantic, and code-pattern.
  arXiv  Detail & Related papers  (2026-05-18T14:48:44Z)
• Guiding Diffusion Models with Semantically Degraded Conditions [19.061619300086875]
  We propose Condition-Degradation Guidance (CDG)<n>CDG replaces the null prompt with a strategically degraded condition, $boldsymbolc_textdeg$.<n>As a lightweight, plug-and-play module, CDG markedly improves compositional accuracy and text-image alignment.
  arXiv  Detail & Related papers  (2026-03-11T13:54:35Z)
• SABER: Small Actions, Big Errors -- Safeguarding Mutating Steps in LLM Agents [52.20768003832476]
  We analyze execution traces on $$-Bench (Airline/Retail) and SWE-Bench Verified.<n>We formalize emphdecisive deviations, earliest action, level divergences that flip success to failure.<n>We introduce cm, a model-agnostic, gradient-free, test-time safeguard.
  arXiv  Detail & Related papers  (2025-11-26T01:28:22Z)
• CLEVER: A Curated Benchmark for Formally Verified Code Generation [53.5486188696892]
  $rm Csmall LEVER$ is a high-quality, curated benchmark of 161 problems for end-to-end verified code generation in Lean.<n>Each problem consists of (1) the task of generating a specification that matches a held-out ground-truth specification, and (2) the task of generating a Lean implementation that provably satisfies this specification.
  arXiv  Detail & Related papers  (2025-05-20T05:15:47Z)
• Language Models May Verbatim Complete Text They Were Not Explicitly Trained On [97.3414396208613]
  We show that a $n$-gram based membership definition can be effectively gamed.<n>We show that it is difficult to find a single viable choice of $n$ for membership definitions.<n>Our findings highlight the inadequacy of $n$-gram membership, suggesting membership definitions fail to account for auxiliary information.
  arXiv  Detail & Related papers  (2025-03-21T19:57:04Z)
• On Distributed Differential Privacy and Counting Distinct Elements [52.701425652208734]
  We study the setup where each of $n$ users holds an element from a discrete set. The goal is to count the number of distinct elements across all users.
  arXiv  Detail & Related papers  (2020-09-21T04:13:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.