Skip to content

feat: Implement Agent Skills MVP#183

Open
creative-CLAi wants to merge 6 commits intolevante-hub:developfrom
creative-CLAi:feature/agent-skills-mvp
Open

feat: Implement Agent Skills MVP#183
creative-CLAi wants to merge 6 commits intolevante-hub:developfrom
creative-CLAi:feature/agent-skills-mvp

Conversation

@creative-CLAi
Copy link

Summary

Implements issue #155 - Agent Skills system following the agentskills.io standard.

What's Included

Core Services

  • skillsService - Discovery, loading, enable/disable of skills
  • IPC handlers - Full API for renderer communication
  • System prompt integration - Enabled skills injected into AI context

UI

  • SkillsPage - Grid view of all discovered skills
  • SkillCard - Toggle component with metadata display
  • Sidebar item - Quick access from navigation

Features

  • 📁 Auto-creates ~/levante/skills/ directory
  • 🔍 Auto-discovers skills with SKILL.md files
  • 🔄 Refresh button to re-scan directory
  • ✅ Toggle skills on/off
  • 💾 Persists enabled state in skills-config.json
  • 🌍 i18n support (en/es)

Skill Format

Create a folder in ~/levante/skills/ with a SKILL.md file:

---
name: my-skill
description: What this skill does and when to use it
version: 1.0.0
author: Your Name
category: development
tags: [nodejs, testing]
---

# Skill Instructions

Markdown content that gets injected into the AI's system prompt.

Testing

  1. Run pnpm dev
  2. Go to Skills page (⚡ icon in sidebar)
  3. Open skills folder and create a test skill
  4. Click Refresh
  5. Toggle the skill on
  6. Start a new chat - the skill context should be active

Screenshots

To be added

Dependencies Added

  • gray-matter - YAML frontmatter parsing

Closes #155

cc @olivermontes

olivermontes and others added 3 commits December 30, 2025 13:11
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements issue levante-hub#155 - Agent Skills system following agentskills.io standard.

## Core Implementation

### Main Process
- Add skillsService with discovery, loading, enable/disable
- Add IPC handlers for all skills operations
- Integrate skills context into system prompt builder
- Auto-create ~/levante/skills/ directory on startup

### Types
- Add Skill, SkillMetadata, SkillsConfig types
- Add SkillValidationResult for validation

### Preload
- Add skillsApi with list, get, enable, disable, toggle, refresh
- Expose skills path getter

### Renderer
- Add skillsStore (Zustand) for state management
- Add SkillsPage with grid view of skills
- Add SkillCard component with toggle switch
- Add Skills item to sidebar navigation
- Add en/es translations for skills UI

### Dependencies
- Add gray-matter for YAML frontmatter parsing

## How It Works
1. Skills are folders in ~/levante/skills/ with a SKILL.md file
2. SKILL.md uses YAML frontmatter for metadata (name, description, etc)
3. Enabled skills are injected into the AI system prompt
4. Toggle skills on/off from the Skills page

## Skill Format (SKILL.md)
```yaml
---
name: my-skill
description: What this skill does
version: 1.0.0
author: Author Name
---

# Skill Instructions

Markdown content that gets injected into the AI context.
```

Closes levante-hub#155
@creative-CLAi
Copy link
Author

@olivermontes Ready for review! 🚀

MVP implementation of Agent Skills - ~1100 lines of code for the full feature including UI, services, and i18n.

To test:

  1. pnpm dev
  2. Click ⚡ Skills in sidebar
  3. Open folder → create skill folder with SKILL.md
  4. Refresh → toggle on → chat with skill context active

Ubuntu added 2 commits January 27, 2026 22:07
- getBaseDir() instead of getBasePath()
- writeFile from fs/promises instead of directoryService.writeFile()
@creative-CLAi creative-CLAi changed the base branch from main to develop January 27, 2026 22:27
Resolve pnpm-lock.yaml conflict by regenerating lockfile
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Implementar Agent Skills System

2 participants