HN · GitHub Signals

OPEN SOURCE, IN FOCUS

claw-swe-bench

opensquilla/claw-swe-bench

Unified adapter framework for evaluating agent harnesses (claws) on SWE-bench

Explore on GitHub ↗

Saved projects stay in this browser. No account needed.

TAKE A CLOSER LOOK

Start with the source.

The repository is the place to check installation steps, supported platforms, the license, and current maintenance before choosing a tool.

Read the documentation on GitHub ↗

Keep exploring

Find projects in the same areas, or search the archive for a different approach.

ANOTHER GOOD FIND

Explore related projects

View the archive ↗

Connected by topics and descriptions.

china-qijizhifeng

agentic-harness-engineering

Official code for AHE — Agentic Harness Engineering: observability-driven automatic evolution of coding-agent harnesses (concurrent work with meta-harness). Lifts Terminal-Bench 2 pass@1 69.7%→77.0% over 10 iterations, beats Codex/ACE/Training Free GRPO; frozen harness transfers to SWE-bench-verified and 3 other base models.

Take a look →
jeikeilim

tenet

Cross-platform AI agent harness for 12+ hour autonomous development cycles. Spec-driven, DAG-orchestrated, with a 3-critic evaluation pipeline. Works with Claude Code, OpenCode, and Codex.

Take a look →
10xchengtu

harness-engineering

Set up and improve harness engineering (AGENTS.md, docs/, lint rules, eval systems, project-level prompt engineering) for AI-agent-friendly codebases. Triggers on: new/empty project setup for AI agents, AGENTS.md or CLAUDE.md creation, harness engineering questions, making agents work better on a codebase.

Take a look →