swe-bench repositories
Trending repositories tagged swe-bench, ranked by trending score. Updated daily.
-
1
TIGER-AI-Lab/WebsiteBench
Can AI agents rebuild websites through browser exploration?
GitHub · Python · 575 stars · 49 forks · +32 stars in 24 hours · Trending score 3.12
-
2
Valen-Team/Roller
Turn agent runs into training data. Roller rolls out Claude Code, Codex, OpenClaw and four other Harnesses in disposable Docker containers, grades benchmark tasks in place, and exports SFT-ready trajectories that keep reasoning, images and compacted-context segments.
GitHub · Python · 113 stars · 0 forks · +25 stars in 24 hours · Trending score 1.59
-
3
mahmoudsajjadi/awesome-ai-coding-tools
A curated catalog of verified AI coding assistants, autonomous agents, open-weight models, and MCP servers with benchmarks and decision guides.
GitHub · Python · 19 stars · 17 forks · +1 stars in 24 hours · Trending score 0.88
-
4
CommonstackAI/TwinRouterBench
Per-step LLM routing benchmark with 970 static labels, live SWE-bench evaluation, an open data pipeline, and a public leaderboard.
GitHub · Python · 35 stars · 4 forks · 0 stars in 24 hours · Trending score 0.73
-
5
huggingface/Repo2RLEnv
Turn any repository into verifiable RL environments for coding agents - Harbor tasks you can train on, evaluate and share on the Hugging Face Hub
GitHub · Python · 788 stars · 128 forks · 0 stars in 24 hours · Trending score 0.65
-
6
SE-Emam/EMO-X
EMO-X: adaptive, execution-based evaluation OS for AI agents.
GitHub · Python · 9 stars · 0 forks · +1 stars in 24 hours · Trending score 0.46
-
7
CambrianTech/continuum
Local-first AI citizens on your own hardware. Teams of continuously-learning personas (Rust core, llama.cpp fork, LoRA genomes) that collaborate, remember, dream, and improve on public benchmarks — on a MacBook. Not assistants you rent: minds you host.
GitHub · Rust · 39 stars · 11 forks · 0 stars in 24 hours · Trending score 0.33
-
8
BrandeisPatrick/openbench
Reverse-engineer the RL reward an LLM was trained on — from black-box agentic coding behavior alone. Tested on Opus 4.8, Fable 5, GPT-5.5.
GitHub · Python · 9 stars · 0 forks · 0 stars in 24 hours · Trending score 0.22
-
9
unarbos/arbos
Arbos is a coding agent for one project. Talk to Main, watch workers, keep the work in one place.
GitHub · Rust · 33 stars · 12 forks · 0 stars in 24 hours · Trending score 0.20
-
10
DeepGym/deepgym
RL environments with verifiable rewards you can trust. Sandboxed coding and computer-use tasks, verifiers kept apart from the answer, a QA gate against reward hacking, and infra failures kept out of training. Works with verl, TRL, OpenRLHF and Axolotl.
GitHub · Python · 22 stars · 5 forks · −3 stars in 24 hours · Trending score 0.18
-
11
mupt-ai/self-bench
Benchmark coding agents and models on your own repository with private Harbor-formatted evals built from its merged pull requests.
GitHub · TypeScript · 34 stars · 1 forks · 0 stars in 24 hours · Trending score 0.10
-
12
bruce601080102/ccc-mcp
Stop trusting "tests passed." Orchestrate ChatGPT, Codex & Claude Code over MCP, with evidence for every handoff: host-run tests, versioned shared context, quota fallback and recoverable tasks.
GitHub · Python · 5 stars · 0 forks · 0 stars in 24 hours · Trending score 0.09
-
13
SourceShift/mini-ork
Prove an AI-written fix actually fixes the bug. mini-ork certify reproduces the bug against your repo's own code, attacks the patch with generated invariants in a Docker sandbox, and returns PROVEN / REFUTED / UNVERIFIED with a JSON certificate. Part of a verifier-gated, multi-model agent runtime.
GitHub · Python · 35 stars · 11 forks · 0 stars in 24 hours · Trending score 0.08
-
14
La5u/Parsimony
Measures how much code AI coding agents add, from patches benchmarks already publish
GitHub · Python · 14 stars · 3 forks · 0 stars in 24 hours · Trending score 0.08
-
15
xianyu-sheng/Xenon
Verifiable AI coding agent harness / 可验证的 AI 编码代理:evidence hash-chain · hollow-answer gate · manual verification checklist · single-loop kernel + bidirectional MCP · multi-model · Windows/CJK · listed in awesome-opensource-ai & Collabnix awesome-mcp-lists
GitHub · Python · 56 stars · 11 forks · 0 stars in 24 hours · Trending score 0.07
-
16
hwfengcs/DM-Code-Agent
An inspectable coding agent. Trace changes, diagnose failed runs, and share test evidence. Local-first Python, no-key demo, standalone HTML reports.
GitHub · Python · 152 stars · 15 forks · 0 stars in 24 hours · Trending score 0.06
-
17
facebookresearch/swe-sweep
How many bugs can LMs find & fix in large codebases?
GitHub · Python · 230 stars · 28 forks · 0 stars in 24 hours · Trending score 0.04
-
18
vishal-dehurdle/state-harness
Runtime safety net for LLM agents. Detects token spirals, kills doomed tasks early, tells you exactly why. Rust core, Python SDK. pip install state-harness
GitHub · Python · 18 stars · 2 forks · 0 stars in 24 hours · Trending score 0.03
-
19
BichengWang/multi-agents
Multi-agent teams for real work: run teams of coding agents (Claude Code, Codex, any model) in isolated git worktrees, and measure whether the team beats a solo agent on cost and results.
GitHub · Python · 158 stars · 12 forks · 0 stars in 24 hours · Trending score 0.01
-
20
wojciechwiesner/jit-context
Architectural Moat: JIT-JEV Context OS — Epistemic runtime & JEV System 1 context gate for AI agents (L0 SQLite WAL <3ms, Epistemic Invariants I1–I10, CERN Zenodo DOI: 10.5281/zenodo.22649542)
GitHub · Python · 9 stars · 1 forks · 0 stars in 24 hours · Trending score 0.01
-
21
shreyash-sharma/provenant
Microsoft Build AI 2026 National 1st Place. Local codebase memory for AI coding agents: 63.8% Coverage@5, 75.2% Coverage@10, 60-65x lower context.
GitHub · Python · 16 stars · 1 forks · 0 stars in 24 hours · Trending score 0.00
-
22
vieval-dev/vieval
🧪💥 Evaluation framework based on Vitest, the testing framework you familiar with, for agents, models, and more.
GitHub · TypeScript · 71 stars · 0 forks · 0 stars in 24 hours · Trending score 0.00
-
23
wbopan/retro-harness
RHO: Evolving Agents in the Dark — Retrospective Harness Optimization via Self-Preference. Improving LLM agents from unlabeled past trajectories (arXiv:2606.05922).
GitHub · Python · 68 stars · 8 forks · 0 stars in 24 hours · Trending score 0.00
-
24
adrienbrault/qwen3.8-27b-rtx5090
Qwen3.8-27B on 2× RTX 5090 · 1.49M KV · code/prose t/s: c1 264/216, c16 2,649/2,240 · NVFP4 + vLLM + MTP
GitHub · Shell · 25 stars · 4 forks · 0 stars in 24 hours · Trending score 0.00