AI Paper Atlas

Research map / Large language models / Agents, code and math

Gistify: Codebase-Level Understanding via Runtime Execution

Hyunji Lee, Minseon Kim, Chinmay Singh, Matheus Pereira, Atharv Sonwane, Isadora White, Elias Stengel-Eskin, Mohit Bansal, Zhengyan Shi, Alessandro Sordoni, Marc-Alexandre Côté, Xingdi Yuan and 1 more

ICLR 2026 · Large language models › Agents, code and math · approach: Software engineering agents

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
prompt the model to trace execution (ablation)SWE-agent Claude-4 · Gistify (50 pylint tests, SWE-agent Claude-4)execution fidelity: 42.0 -> 48.0Improved
prompt with explicit reading strategy (ablation)SWE-agent Claude-4 · Gistify (50 pylint tests, SWE-agent Claude-4)execution fidelity: 42.0 -> 50.0Improved
add RepoGraph tool (ablation)SWE-agent Claude-4 · Gistify (50 pylint tests, SWE-agent Claude-4)execution fidelity: 42.0 -> 52.0Improved
add gold execution tracing tool (ablation)SWE-agent Claude-4 · Gistify (50 pylint tests, SWE-agent Claude-4)execution fidelity: 42.0 -> 56.0Improved
give full bash execution access (ablation)SWE-agent Claude-4 · Gistify (50 pylint tests, SWE-agent Claude-4)execution fidelity: 42.0 -> 52.0Improved
restrict execution to edit-and-execute instead of full bash (ablation)SWE-agent Claude-4 · Gistify (50 pylint tests, SWE-agent Claude-4)execution fidelity: 52.0 -> 56.0Improved
provide gold test function in prompt (ablation)SWE-agent Claude-4 · Gistify (pylint)execution fidelity: 42.0 -> 60.0Improved
enable execution tools in main setting (ablation)3 frameworks x 4 models · Gistifyexecution fidelity: e.g. mini-SWE-agent GPT-5-mini 17.1 -> 24.0; Copilot GPT-5-mini 58.0 -> 55.3Mixed
use agentic multi-step model instead of static LLM with gold files (ablation)Claude-4 · Gistify subsetexecution fidelityImproved
select hardest tests by trace length and files (Gistify-hard)Claude-4 · Gistify-hardexecution fidelity: 43 -> 21Hurt
static LLM given gold files vs agenticstatic coding LLM · Gistifyline existenceImproved
move from bash-only mini-SWE-agent to Copilot toolsetGPT-5-mini · Gistifyexecution fidelity: 17.1 -> 58.0Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy