
trae-community/trae-skills
Developer toolsЭта страница переведена автоматически. Технические детали смотрите в оригинале. Открыть оригинал
AI coding execution control skills | AI 编程执行约束技能
Обзор
1. Clone this repository. 2. Put skills into the right location for TRAE: - Project skills: .trae/skills/ /SKILL.md - Global skills: ~/.trae/skills/ /SKILL.md 3. Refresh TRAE's skill discovery in settings (location varies by TRAE version). 4. Ask TRAE for a task that matches a skill’s description, for example: - “Use the git-commit-generator skill to draft a commit message for the current changes.” - “Use the cn-punctuation-checker skill to check Chinese punctuation in this document.” Agent Skills are folders of instructions, scripts, and resources that an AI agent can discover and load dynamically to perform specialized tasks in a repeatable way. A typical skill is a directory with a SKILL.md file that contains: - YAML frontmatter metadata (especially name and description) - A Markdown body with steps, guidelines, and examples This pattern keeps the agent’s core rules lightweight while making SOP-style workflows portable and shareable. - Skills vs.
README
Показан фрагмент оригинального README. Открыть оригинал.
TRAE Agent Skills
Community-maintained Agent Skills for TRAE.
Quickstart
- Clone this repository.
- Put skills into the right location for TRAE:
- Project skills:
.trae/skills//SKILL.md - Global skills:
~/.trae/skills//SKILL.md
- Project skills:
- Refresh TRAE’s skill discovery in settings (location varies by TRAE version).
- Ask TRAE for a task that matches a skill’s description, for example:
- “Use the git-commit-generator skill to draft a commit message for the current changes.”
- “Use the cn-punctuation-checker skill to check Chinese punctuation in this document.”
What are Agent Skills?
Agent Skills are folders of instructions, scripts, and resources that an AI agent can discover and load dynamically to perform specialized tasks in a repeatable way. A typical skill is a directory with a SKILL.md file that contains:
- YAML frontmatter metadata (especially
nameanddescription) - A Markdown body with steps, guidelines, and examples
This pattern keeps the agent’s core rules lightweight while making SOP-style workflows portable and shareable.
Skills vs. other features in TRAE
- Skills vs. Rules: rules are fully injected into every chat and continuously occupy context; skills are loaded on-demand only when called.
- Skills vs. MCP servers: skills describe how TRAE should accomplish a task; MCP servers provide tools that TRAE can call. For example, a Playwright MCP server provides browser automation tools, while a testing skill defines test structure, conventions, and execution workflow.
Skill types in TRAE
- Global skills: reusable across projects (personal/team conventions, general toolchain workflows, long-term output preferences).
- Project skills: apply only to the current project (project-specific business rules, architecture constraints, project scaffolding/testing workflows).
Repository layout
This repository is intended to follow a simple, discoverable layout:
skills/
_template/ # Template for creating new skills
SKILL.md
/
SKILL.md # (Mandatory) Core instructions for the agent
(optional) examples/ # Input/output examples
input.md
output.md
(optional) templates/ # Reusable templates
component.tsx
(optional) resources/ # Reference files, scripts, or assets
style-guide.md
To create a skill, start with the skill template and follow the contributing guide.
Skills catalog
| Skill | Description | Usage Scenario | Status |
|---|
Рекомендуемые инструменты
Попробуйте другой запрос или уберите фильтр.
Установка
npx skillfish add trae-community/trae-skills