Git Good Bench Train: AI Agent Trajectories for Git Task Resolution

A subset of 17,469 samples for collecting trajectories of AI agents resolving git tasks. The data was collected from 816 unique, open-source GitHub repositories with permissive licenses and high activity metrics. The dataset was created by JetBrains and last updated on November 18, 2025.

Use Cases

• Training reinforcement learning agents for merge conflict resolution based on the supported scenario.
• Developing models for file-commit chain generation based on the described sample type.
• Benchmarking AI performance on Git operations in Python, Java, and Kotlin based on the supported programming languages.

Strengths

• Contains 17,469 samples for model training purposes.
• Sourced from 816 unique, high-quality GitHub repositories with permissive licenses, >=1000 stars, >=5 branches, and >=10 contributors.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known but total dataset size and file formats are unknown, which may limit suitability assessment.

Provenance

Source

JetBrains via Hugging Face.

Collection Method

Collected from trajectories of AI agents resolving git tasks in open-source GitHub repositories.

Time Range

null

Freshness

Last updated 2025-11-18 09:06:31.

Geography

null