Skills Registry MCP Server

#!/usr/bin/env python3 """ Import skills from GitHub repositories into the Skills Registry """ import os import sys import json import asyncio import tempfile import shutil from pathlib import Path from typing import List, Dict, Any import subprocess # Add parent directory to path for imports sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..', 'src')) from database import Database from search import SearchEngine # Skills to import from GitHub SKILL_SOURCES = [ # Official Anthropic Skills { "repo": "https://github.com/anthropics/skills.git", "paths": [ # Document Skills "document-skills/docx", "document-skills/pdf", "document-skills/pptx", "document-skills/xlsx", "document-skills/markdown", # Dev & Code "code-execution", "code-review", "code-explanation", "test-generation", # Data "data-skills/csv-analysis", "data-skills/json-parser", # Writing & Research "research-assistant", "summarization", # Media "audio-transcription", "image-captioning", # Utility "file-manager", "task-runner", "email-parser", "skill-creator", "template-skill", # Design "canvas-design", "slack-gif-creator", "brand-guidelines", "theme-factory", "artifacts-builder", "mcp-builder", "internal-comms", ], "author": "anthropic", "verified": True, "ai_generated": False, }, # Obra's Superpowers Collection { "repo": "https://github.com/obra/superpowers.git", "paths": [ "skills/using-git-worktrees", "skills/test-driven-development", "skills/subagent-driven-development", "skills/executing-plans", "skills/finishing-a-development-branch", "skills/root-cause-tracing", "skills/writing-plans", "skills/writing-skills", "skills/brainstorming", "skills/using-superpowers", "skills/sharing-skills", "skills/commands", "skills/receiving-code-review", "skills/requesting-code-review", "skills/dispatching-parallel-agents", "skills/defense-in-depth", "skills/systematic-debugging", "skills/testing-anti-patterns", "skills/testing-skills-with-subagents", "skills/verification-before-completion", "skills/condition-based-waiting", ], "author": "obra", "verified": True, "ai_generated": False, }, # Composio Awesome Skills { "repo": "https://github.com/ComposioHQ/awesome-claude-skills.git", "paths": [ "changelog-generator", "content-research-writer", "image-enhancer", "video-downloader", "invoice-organizer", "raffle-winner-picker", ], "author": "composio", "verified": True, "ai_generated": False, }, # Michal Parkola's Tapestry Skills { "repo": "https://github.com/michalparkola/tapestry-skills-for-claude-code.git", "paths": [ "ship-learn-next", "article-extractor", "youtube-transcript", "tapestry", ], "author": "michalparkola", "verified": True, "ai_generated": False, }, # mhattingpete's Skills Marketplace { "repo": "https://github.com/mhattingpete/claude-skills-marketplace.git", "paths": [ "git-pushing", "review-implementing", "test-fixing", ], "author": "mhattingpete", "verified": False, "ai_generated": False, }, # coffeefuelbump's CSV Summarizer { "repo": "https://github.com/coffeefuelbump/csv-data-summarizer-claude-skill.git", "paths": ["."], "author": "coffeefuelbump", "verified": False, "ai_generated": False, }, # Community Skills (single repos) { "repo": "https://github.com/smerchek/claude-epub-skill.git", "paths": ["."], "author": "smerchek", "verified": False, "ai_generated": False, }, { "repo": "https://github.com/jthack/ffuf_claude_skill.git", "paths": ["."], "author": "jthack", "verified": False, "ai_generated": False, }, ] def clone_repo(repo_url: str, target_dir: str) -> bool: """Clone a git repository""" try: print(f"Cloning {repo_url}...") subprocess.run( ["git", "clone", "--depth", "1", repo_url, target_dir], check=True, capture_output=True ) return True except subprocess.CalledProcessError as e: print(f"Failed to clone {repo_url}: {e.stderr.decode()}") return False def find_skill_md(skill_path: Path) -> Path: """Find SKILL.md file in the skill directory""" # Try exact match first skill_md = skill_path / "SKILL.md" if skill_md.exists(): return skill_md # Try case-insensitive search for file in skill_path.glob("*.md"): if file.name.upper() == "SKILL.MD": return file # Try README.md as fallback readme = skill_path / "README.md" if readme.exists(): return readme return None def extract_metadata_from_skill(skill_md_path: Path, skill_md_content: str) -> Dict[str, Any]: """Extract metadata from SKILL.md content""" lines = skill_md_content.split('\n') # Try to find title (first # heading) name = skill_md_path.parent.name.replace('-', ' ').title() for line in lines[:20]: # Check first 20 lines if line.startswith('# '): name = line[2:].strip() break # Try to find description (first paragraph after title) description = "" found_title = False for line in lines[:50]: # Check first 50 lines if line.startswith('# '): found_title = True continue if found_title and line.strip() and not line.startswith('#'): description = line.strip() break # Infer category from path parent_name = skill_md_path.parent.name.lower() if 'doc' in parent_name or 'pdf' in parent_name or 'pptx' in parent_name or 'xlsx' in parent_name: category = "documents" elif 'code' in parent_name or 'dev' in parent_name or 'git' in parent_name: category = "development" elif 'design' in parent_name or 'art' in parent_name or 'canvas' in parent_name: category = "design" elif 'data' in parent_name or 'csv' in parent_name: category = "data" elif 'research' in parent_name or 'article' in parent_name: category = "research" elif 'test' in parent_name or 'debug' in parent_name: category = "testing" else: category = "general" # Extract tags from content tags = [] content_lower = skill_md_content.lower() if 'python' in content_lower: tags.append('python') if 'javascript' in content_lower or 'react' in content_lower: tags.append('javascript') if 'git' in content_lower: tags.append('git') if 'test' in content_lower: tags.append('testing') return { "name": name, "description": description or f"Skill: {name}", "category": category, "tags": tags, } async def import_skills(): """Import all skills from GitHub repositories""" # Initialize database db_url = os.getenv("DATABASE_URL", "postgresql://skills:skills_dev_password@localhost:5432/skills_registry") db = Database(db_url) # Initialize search engine openai_key = os.getenv("OPENAI_API_KEY") anthropic_key = os.getenv("ANTHROPIC_API_KEY") search = SearchEngine(db, openai_key, anthropic_key) total_imported = 0 total_failed = 0 # Create temporary directory for cloning repos with tempfile.TemporaryDirectory() as temp_dir: temp_path = Path(temp_dir) for source in SKILL_SOURCES: repo_url = source["repo"] repo_name = repo_url.split('/')[-1].replace('.git', '') clone_path = temp_path / repo_name # Clone repository if not clone_repo(repo_url, str(clone_path)): print(f"Skipping {repo_url} due to clone failure") continue # Import each skill from this repo for skill_path_str in source["paths"]: skill_path = clone_path / skill_path_str if not skill_path.exists(): print(f"⚠️ Path not found: {skill_path}") total_failed += 1 continue # Find SKILL.md skill_md_path = find_skill_md(skill_path) if not skill_md_path: print(f"⚠️ No SKILL.md found in: {skill_path}") total_failed += 1 continue # Read skill content try: with open(skill_md_path, 'r', encoding='utf-8') as f: skill_md_content = f.read() except Exception as e: print(f"⚠️ Failed to read {skill_md_path}: {e}") total_failed += 1 continue # Extract metadata metadata = extract_metadata_from_skill(skill_md_path, skill_md_content) # Create skill in database try: skill_id = await db.create_skill( name=metadata["name"], description=metadata["description"], skill_md_content=skill_md_content, category=metadata["category"], tags=metadata["tags"], author_id=source["author"], visibility="public", ai_generated=source["ai_generated"] ) # Generate embedding await search.generate_embedding(skill_id) # Mark as verified if from trusted source if source["verified"]: conn = db._get_conn() try: with conn.cursor() as cur: cur.execute( "UPDATE skills SET verified = TRUE WHERE skill_id = %s", (skill_id,) ) conn.commit() finally: conn.close() print(f"✅ Imported: {metadata['name']} ({skill_id})") total_imported += 1 except Exception as e: print(f"⚠️ Failed to import {metadata['name']}: {e}") total_failed += 1 print("\n" + "="*60) print(f"Import complete!") print(f"✅ Successfully imported: {total_imported}") print(f"⚠️ Failed: {total_failed}") print("="*60) if __name__ == "__main__": asyncio.run(import_skills())