skillsbench
Evaluate the effectiveness and efficiency of AI agents using skills.
- Type
- Agent Skill
- Open source
- Yes
- GitHub Stars
- ★ 1.4k
- Source
- skill-github
- Repository
- github.com/benchflow-ai/skillsbench
Overview
SkillsBench is a benchmarking tool designed to evaluate how effectively AI agents utilize skills. It employs a gym-like benchmarking approach to measure an agent's skill utilization when executing specific workflows. The tool aims to build the most comprehensive and high-quality benchmark for agent skills, featuring tasks that require the combination of multiple skills. It supports major models such as GPT-5.5, Claude Opus 4.8, and others. Users can get started by installing the BenchFlow CLI and running the relevant commands.
Capabilities
- ▪Skill effectiveness evaluation
- ▪Agent behavior evaluation
- ▪Multi-skill composition task design
- ▪Support for multiple mainstream models
Use cases
Setup
git clone https://github.com/benchflow-ai/skillsbench.git; cd skillsbench; uv tool install benchflow; uv sync --locked
This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.
FAQ
How do I get started with SkillsBench?
Clone the repository, install the BenchFlow CLI, sync the tools, then run the task check.
What API keys are required?
API keys from providers such as Anthropic and OpenAI are required.
Related skills
gemini-cli
Gemini CLI is an open-source AI agent that brings Gemini's powerful capabilities directly into the terminal.
private-gpt
Provides a complete API layer for local models, supporting RAG, skills, tools, and more.
planning-with-files
File-based persistent planning skill, ideal for AI coding agents and long-running tasks.
scientific-agent-skills
Transform any AI agent into a scientific assistant with 147 ready-to-use research skills.
susi_alexa_skill
A skill that enables question-and-answer interactions between Alexa and Susi AI.
claude-context
Provides code search capabilities for Claude Code, turning the entire codebase into context.