Claude Skill
huggingface/upskill
A framework for systematically generating and evaluating skills for AI code agents. Enhances capabilities in code understanding, generation, optimization, and debugging with comprehensive metrics.
Overview
Repository
Install this Skill
uvx upskillRegistry
Summary
Upskill is a framework for generating and evaluating specialized skills for code agents like Claude Code, Open Code, and OpenAI Codex. It provides a systematic approach to skill development across code understanding, generation, optimization, debugging, and documentation, paired with a comprehensive evaluation system covering functionality, quality, efficiency, and practicality.
为代码代理（如Claude Code、Open Code、OpenAI Codex）生成并评估技能集 **技能生成方向：** 1. **代码理解与解析** - 多语言语法分析（Python、JavaScript、Java等） - 代码结构识别（函数、类、模块关系） - 依赖关系图谱构建 2. **智能代码生成** - 根据自然语言描述生成功能代码 - 代码补全与片段建议 - 测试用例自动生成 3. **代码优化与重构** - 性能瓶颈检测与优化建议 - 代码风格规范化 - 架构模式推荐 4. **调试与错误处理** - 异常原因分析 - 堆栈跟踪解读 - 修复方案推荐 5. **文档与知识管理** - 代码注释自动生成 - API文档生成 - 技术决策记录 **评估指标体系：** 1. **功能性指标** - 代码正确率（编译/运行通过率） - 功能实现完整度 - 边界情况处理能力 2. **质量指标** - 代码可读性评分 - 算法复杂度合理性 - 安全漏洞检测率 3. **效率指标** - 响应延迟（毫秒级） - 上下文理解准确率 - 多轮对话保持一致性 4. **实用性指标** - 开发效率提升幅度 - 学习曲线平缓度 - 集成开发环境适配性 **评估方法：** - 基准测试集（HumanEval、MBPP等） - A/B测试对比人类开发者 - 用户满意度调研 - 长期使用留存率分析 **持续优化机制：** 1. 实时反馈闭环系统 2. 技能模块热更新 3. 领域自适应训练 4. 多代理协作验证 通过系统化技能设计和多维评估，可构建具备专业级编码能力的智能代理，同时需注意平衡自动化程度与开发者控制权，建立人机协同的最佳实践范式。
Key features
- Systematic skill generation for code agents
- Multi-dimensional evaluation metrics (functionality, quality, efficiency, practicality)
- Focus on code understanding, generation, optimization, and debugging
- Continuous optimization mechanism with real-time feedback
- Integration with benchmark test sets (e.g., HumanEval, MBPP)
Use cases
- Enhancing code agent capabilities for software development
- Benchmarking and comparing different code generation models
- Automating code review, optimization, and documentation tasks
- Training and evaluating AI-powered coding assistants
- Research on AI code generation and agent performance
README excerpt
<img width="1920" height="1080" alt="upskill_banner" src="https://github.com/user-attachments/assets/b71fd417-7d23-4f5d-aa89-06ea6b284d1b" /> # UPskill Generate and evaluate agent skills based on traces with agents. Create skills with teacher models (expensive/slow) that student models (cheap/fast) can use to perform harder tasks reliably. > [!TIP] > > UPskill v2 - recommended default config file now runs evaluations on Hugging Face Jobs. Make sure > to set your `HF_TOKEN` and use `--artifact-repo <dataset-name>` for job creation and result capture ## Quick Start Install upskill: ```bash uv pip install upskill # or just use uv uvx upskill ``` Create a new skill ```bash upskill generate "write good git commit messages" # or based on previous agent traces upskill generate "document the pattern" --from ./trace.md # Skills are saved to ./skills/{skill-name}/ by default ``` Generate a skill with a teaching model and evaluate it on a student model. ```bash upskill generate "write good git commit messages" --model sonnet --eval-model haiku ``` Benchmark a set of models against a skill. ```bash upskill eval ./skills/git-commit-messages/ -m haiku -m sonnet # logs pretty printed to the terminal ``` View the results later. ```bash upskill runs --skill git-commit-messages ``` ## Development checks This repo uses a CI flow inspired by `fast-agent` with separate format, lint, typecheck, and test stages. Install dev dependencies: ```bash uv sync --extra dev ``` Run the quality gates locally: ```bash uv run scripts/format.py uv run scripts/lint.py uv run scripts/typecheck.py uv run scripts/cpd.py --check uv run --extra dev pytest -v ``` Or use the helper script to run the whole sequence: ```bash uv run scripts/check.py ``` Add `--sync` to include `uv sync --
Topics
No topics yet.