SkillsBench
SkillsBench evaluates how well skills work and how effective agents are at using them.
About
The first benchmark for evaluating AI agent skills. 87 tasks, 24 model-harness configurations, 3 trials per task. See how skills improve agent performance across diverse domains.
Open Source Health
- Stars
- 1,768
- Forks
- 367
- License
- Apache-2.0
- Last commit
- 3 months ago
Resources & Links
Related Categories
Vendor
BenchFlow
Open-source project on GitHub: SkillsBench
Quick Links
Open Source
Related Products
Investorskills
Questflow: AI Finance Agent
Top category match
Math Modeling Skill
数学建模技能 - 面向 CUMCM、MCM/ICM 等数学建模竞赛的三阶段工作流：建模分析、Python/MATLAB 编程与 DOCX 论文生成。包含丰富的算法资源库(优化/预测/评价/图论/机器学习等)、角色指导文档、论文模板和实用工具脚本
Top category match
Cc Thinking Skills
28 eval-informed mental models and critical-thinking skills for Claude Code, GitHub Copilot, Codex, Cursor, and other Agent Skills-compatible tools
Top category match
Audit Skills
专注于代码审计skill，最小化轻量化skill，只负责安全边界。
Top category match
Dotnet Skills
Claude Code skills and sub-agents for .NET Developers
Top category match