EvoSkill

EvoSkill

EvoSkill is an open-source framework that automatically discovers and synthesizes reusable agent skills to enhance coding agent performance.

Agent SkillDevOpen source
Type
Agent Skill
Open source
Yes
GitHub Stars
★ 1.0k
Source
skill-github

Overview

EvoSkill is an open-source framework that automatically discovers and synthesizes reusable agent skills from failed trajectories, improving coding agent performance. It proposes multiple skill and prompt variations, evaluates new variants on retained data, and generates entirely new agent programs in each iteration. EvoSkill supports various coding agents such as Claude Code, Codex CLI, OpenCode, and can work with any model provider and model. Install in seconds, then run `evoskill init` and `evoskill run` to add AI-generated skills and prompts to any coding agent.

Capabilities

  • Automatic discovery and synthesis of agent skills
  • Support for multiple coding agents
  • Compatibility with various model providers and models
  • Cross-task skill transfer

Use cases

Improving coding agent performanceAutomatically generating and refining AI skillsSkill transfer across agents and models

Setup

Requires: API KeyNode 环境
```bash
github.com/sentient-agi/EvoSkill
```, then run `evoskill init` and `evoskill run`.

This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.

FAQ

Which coding agents does EvoSkill support?

It supports multiple coding agents such as Claude Code, Codex CLI, OpenCode, and others.

Does EvoSkill require a specific model provider?

No, it can work with any model provider and model.

Related skills