Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Git Context Controller: Manage the Context of LLM-based Agents like Git

Junde Wu, Minhao Hu, Jiayuan Zhu, Jiazhen Pan, Yuyuan Liu, Min Xu, Yueming Jin

TL;DR

Git-Context-Controller (GCC) is introduced, a structured context management framework inspired by software version control systems that allows agents to manage long-term goals, recover and transfer reasoning across sessions, and coordinate multi-trajectory problem solving in a principled manner.

Abstract

Large language model (LLM) agents have demonstrated strong capabilities in long-horizon tasks by interleaving reasoning with tool use. However, as these agents scale to complex workflows such as software engineering and open-ended research, context management becomes a fundamental bottleneck: interaction histories grow unbounded, become costly to maintain, and are difficult to reuse across sessions and agents. We introduce \textbf{Git-Context-Controller (GCC)}, a structured context management framework inspired by software version control systems. GCC elevates agent context from a transient token stream to a persistent, navigable memory workspace with explicit operations -- \texttt{COMMIT}, \texttt{BRANCH}, \texttt{MERGE}, and \texttt{CONTEXT}, that enable milestone-based checkpointing, isolated exploration of alternative reasoning paths, and hierarchical retrieval of historical context. By organizing agent memory as a versioned file system, GCC allows agents to manage long-term goals, recover and transfer reasoning across sessions, and coordinate multi-trajectory problem solving in a principled manner. Empirically, agents equipped with GCC achieve state-of-the-art performance on both SWE-Bench and BrowseComp benchmarks. On SWE-Bench Verified, GCC improves task resolution by over 13\% relative to strong long-context baselines and outperforms 26 existing open and commercial systems, reaching over 80\% success rate. The project will be open-sourced for the research community. The algorithm has been incorporated to the project: https://one-context.com

Git Context Controller: Manage the Context of LLM-based Agents like Git

TL;DR

Git-Context-Controller (GCC) is introduced, a structured context management framework inspired by software version control systems that allows agents to manage long-term goals, recover and transfer reasoning across sessions, and coordinate multi-trajectory problem solving in a principled manner.

Abstract

Large language model (LLM) agents have demonstrated strong capabilities in long-horizon tasks by interleaving reasoning with tool use. However, as these agents scale to complex workflows such as software engineering and open-ended research, context management becomes a fundamental bottleneck: interaction histories grow unbounded, become costly to maintain, and are difficult to reuse across sessions and agents. We introduce \textbf{Git-Context-Controller (GCC)}, a structured context management framework inspired by software version control systems. GCC elevates agent context from a transient token stream to a persistent, navigable memory workspace with explicit operations -- \texttt{COMMIT}, \texttt{BRANCH}, \texttt{MERGE}, and \texttt{CONTEXT}, that enable milestone-based checkpointing, isolated exploration of alternative reasoning paths, and hierarchical retrieval of historical context. By organizing agent memory as a versioned file system, GCC allows agents to manage long-term goals, recover and transfer reasoning across sessions, and coordinate multi-trajectory problem solving in a principled manner. Empirically, agents equipped with GCC achieve state-of-the-art performance on both SWE-Bench and BrowseComp benchmarks. On SWE-Bench Verified, GCC improves task resolution by over 13\% relative to strong long-context baselines and outperforms 26 existing open and commercial systems, reaching over 80\% success rate. The project will be open-sourced for the research community. The algorithm has been incorporated to the project: https://one-context.com

Paper Structure

This paper contains 22 sections, 2 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Method
  3. GCC File System
  4. GCC Commands
  5. COMMIT <summary>
  6. BRANCH <name>
  7. MERGE <branch>:
  8. CONTEXT <options>:
  9. Experiment
  10. Datasets
  11. Implementation
  12. Baselines
  13. Experimental Results
  14. GCC surpasses all controlled baselines.
  15. GCC yields even larger gains on SWE-Bench Verified.
  16. ...and 7 more sections

Figures (3)

  • Figure 1: Results on BrowseComp-Plus and SWE-Bench Verified comparing baseline model performance with improvements achieved by equipping models with GCC.
  • Figure 2: Illustration of GCC in action across two workflows: web-search reasoning and software debugging—showing how agents branch, explore, and merge structured context during long-horizon tasks.
  • Figure 3: Number of context retrieval calling, number of branches, the context length and number of RoadMap updates with the increase of the inference steps