Skip to content

Repository files navigation

AI Agent Skills Repository

A collection of reusable skills for AI coding assistants and agents. These skills extend the capabilities of AI agents by providing specialized knowledge, workflows, and executable scripts for specific domains.

🎯 What Are Skills?

Skills are domain-specific extensions that provide:

  • Specialized Knowledge — Expertise in specific domains (e.g., document processing, data analysis)
  • Standardized Workflows — Step-by-step procedures for complex tasks
  • Executable Scripts — Tools and scripts to automate repetitive operations
  • Best Practices — Proven methodologies and patterns

Each skill is self-contained and includes documentation (SKILL.md), scripts, and references needed for the AI agent to perform tasks effectively.

📦 Available Skills

The full, detailed catalogue (features, use cases, triggers, and tech stack for each skill) lives in SKILLS.md.

Skill Type One-line Description
eddx-translate Basic Translate text labels in EdrawMax/EdrawMind (.eddx) diagrams seamlessly
kb-ingest Orchestrator Thin orchestrator that turns .pptx slides into knowledge-base-ready JSONL chunks for RAG / agent pipelines
meeting-minutes Basic Turn a meeting transcript plus a Markdown template into clean, structured meeting minutes
meeting-minutes-export Orchestrator Orchestrator that runs meeting-minutes and exports the result to Word (.docx)
pptx-translate Basic Translate PowerPoint presentations while preserving all formatting
pptx2md Basic Convert .pptx slides to Markdown with images extracted, for editing, search, re-style, or KB ingestion

🚀 How to Use These Skills

For AI Agents (CodeBuddy, Claude, etc.)

  1. Install the skill by copying the skill directory to your agent's skills folder
  2. Load the skill — the agent reads SKILL.md to understand the workflow
  3. Execute tasks — the agent follows the documented procedures and uses provided scripts

For Developers

Each skill directory contains:

skill-name/
├── SKILL.md           # Skill documentation (required)
├── scripts/           # Executable scripts
├── assets/            # Static assets (if any)
└── references/        # Reference materials (if any)

🛠️ Development Tools

pack_and_install_skill.py — Auto-package & Install Skills

A utility script to automatically package skills, install them to WorkBuddy, and manage bidirectional sync.

Features:

  • 📦 Auto-packaging: Packages skill directory into a .zip file
  • 📥 Auto-installation: Unzips the package to WorkBuddy's skills folder
  • 🔒 Respects .gitignore: Excludes all files/directories matched by .gitignore patterns (walking up to the repo root) — in addition, always excludes a top-level tests directory (exact name, case-sensitive)
  • 🔄 Smart overwrite: Automatically removes existing zip package before creating a new one
  • 💾 Backup before install: Can backup existing skills before overwriting (with --backup flag)
  • ☝️ Backup-only mode: Backup existing installed skill without any packaging/installation (with --backup-only flag)
  • 📦 Zip-only mode: Package skill without installing (with --zip-only flag)
  • 🔁 Update local from WorkBuddy: Reverse-sync — copy the installed version from WorkBuddy back to your local source directory, keeping both a WorkBuddy backup and a local backup (with --update-local flag)
  • 🔍 Diff installed vs local: Compare the installed skill in WorkBuddy with your local source, showing files that are added, removed, or modified — automatically ignoring the same files/directories that packaging excludes (with --diff flag)

Usage:

# Full workflow: package + install (default)
python pack_and_install_skill.py <skill_directory>

# Package only, skip installation
python pack_and_install_skill.py <skill_directory> --zip-only

# Install with backup of existing skill
python pack_and_install_skill.py <skill_directory> --backup

# Backup only (no packaging/installation)
python pack_and_install_skill.py <skill_directory> --backup-only

# Sync WorkBuddy installed version back to local source
python pack_and_install_skill.py <skill_directory> --update-local

# Compare installed skill against local source (ignores packaging-excluded files)
python pack_and_install_skill.py <skill_directory> --diff

How it works:

  1. Default mode (package + install):

    • Removes existing zip package if present
    • Packages skill directory into a .zip file (excluding files matched by .gitignore rules, plus a top-level tests directory)
    • Removes existing skill from WorkBuddy if present
    • Unzips the package to WorkBuddy's skills folder
  2. Zip-only mode (--zip-only):

    • Only creates the .zip package
    • Skips installation to WorkBuddy
  3. Backup-only mode (--backup-only):

    • Detects the skill name from the provided directory
    • Checks if a corresponding skill exists in WorkBuddy
    • Copies the installed skill to <skill_name>.backup in WorkBuddy's skills folder
    • No packaging or installation is performed
  4. Update local from WorkBuddy (--update-local):

    • Step 1: Backs up the installed version in WorkBuddy to <skill_name>.backup
    • Step 2: Backs up your local source directory to <skill_name>.local.backup
    • Step 3: Removes the local source directory and copies the WorkBuddy version back
    • Safely synchronizes any changes made directly in WorkBuddy back to your source tree
  5. Diff installed vs local (--diff):

    • Reads the skill name from the provided directory
    • Compares the installed version in WorkBuddy against your local source directory
    • Reports three categories of drift: files only in local (not yet installed), files only in installed (added/changed directly in WorkBuddy), and files modified (same path, different content)
    • Applies the same exclusion rules as packaging (.gitignore plus a top-level tests directory), so irrelevant files never appear as false differences

WorkBuddy Installation Path:

  • Default: C:\Users\%USERNAME%\.workbuddy\skills\
  • Auto-detected from system username

Requirements:

  • Python 3.6+
  • No additional dependencies (uses standard library only)

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

About

Extend AI agents with specialized skills, workflows, and domain expertise.

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages