claw-eval/claw-eval

Claw-Eval is an evaluation harness for evaluating LLM as agents. All tasks verified by humans.

24
/ 100
Experimental

This project helps AI researchers and developers reliably test and compare how well large language models (LLMs) perform complex, real-world tasks. It takes an LLM as input, puts it through various challenges in a sandboxed environment, and then provides a transparent score indicating its performance, verified through human checks. It's designed for anyone building or evaluating AI agents that need to act autonomously.

Use this if you are developing or selecting an AI agent and need an objective, reproducible way to measure its ability to handle practical tasks, from web searches to content creation.

Not ideal if you are looking for a simple API to integrate an LLM into your application without needing to rigorously benchmark its agentic capabilities.

AI-agent-development LLM-evaluation AI-benchmarking agentic-workflow-testing model-comparison
No License No Package No Dependents
Maintenance 10 / 25
Adoption 8 / 25
Maturity 3 / 25
Community 3 / 25

How are scores calculated?

Stars

68

Forks

1

Language

Python

License

Last pushed

Mar 12, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/claw-eval/claw-eval"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.