swe_bench

Automate SWE-bench benchmark execution and validation in Docker containers.

Updated Jun 22, 2026
One-click install
npx skills add https://github.com/alangeb/tau --skill swe-bench-alangeb
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: swe_bench
Source: https://github.com/alangeb/tau/tree/main/src/skills/swe_bench
Command: npx skills add https://github.com/alangeb/tau --skill swe-bench-alangeb

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill streamlines the complex, multi-step process of running SWE-bench evaluations, managing containerized environments, and analyzing agent-generated patches.

Core Features & Use Cases

  • Automated Pipeline: Orchestrates the full lifecycle from container preparation and agent execution to patch extraction and test evaluation.
  • Artifact Management: Provides structured logging and audit trails for every test run, including stdout, context files, and patch diffs.
  • Use Case: Use this to run a batch of GitHub issues through an AI agent, automatically verify the generated fixes against the project test suite, and generate a summary report of pass/fail results.

Quick Start

Use the swe_bench skill to start the evaluation pipeline for the next five test cases using the cuda llm group.

Frequently Asked Questions about swe_bench

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate SWE-bench evaluation for AI-generated code patches?

To automate SWE-bench evaluation, this Skill orchestrates the full lifecycle from container preparation and agent execution to patch extraction and test validation against existing project test harnesses.

Do I need Docker to run SWE-bench patch testing workflows?

Yes, you need a configured Docker environment to run SWE-bench patch testing workflows, as the Skill executes and validates software engineering benchmarks within isolated container environments.

How does automated benchmark validation handle test artifacts and logs?

Automated benchmark validation provides structured logging and artifact management for every test run, capturing stdout, context files, and patch diffs to generate a comprehensive audit trail.

Can I batch process GitHub issues through an AI agent for automatic fix verification?

You can batch process GitHub issues through an AI agent to automatically verify generated fixes against the project test suite and generate a summary report of pass/fail results.

What is the tau agent framework's role in software engineering benchmark execution?

The tau agent framework manages test lifecycles and artifact generation during software engineering benchmark execution, requiring configured access to automate the end-to-end evaluation pipeline.