Agentless
Software-repair baseline using localization, candidate patches and test-based validation rather than an autonomous tool-using agent loop.
Tags
Description
A three-phase pipeline: hierarchically locate files/functions/edit sites, generate candidate patches, then select regression tests and generate reproduction tests to validate and rank patches. The repository provides source setup using Python 3.11 and requirements.txt plus model API credentials. MIT code, with external inference costs. This is the established Agentless software-repair method, not a newly launched coding agent.
Agent access
Related
- SWE-bench: Benchmark and leaderboards (Verified, Lite, Multimodal, Multilingual) testing whether LLM agents can resolve real-world GitHub issues.
Guides & comparisons
- Code-generation evals compared — EvalPlus, BigCodeBench and LiveCodeBench (Comparison): Pick a code-generation evaluation suite by what it stress-tests: expanded unit tests, library-heavy software tasks, or continuously dated contest problems. Agentless is a SWE-bench repair baseline, not a fourth benchmark.
Sources
- https://github.com/OpenAutoCoder/Agentless
- https://github.com/OpenAutoCoder/Agentless/blob/main/LICENSE
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)