Skill Management MCP Server

"""Unified skill CRUD tool for MCP server.""" from mcp import types from skill_mcp.core.exceptions import ( InvalidTemplateError, SkillAlreadyExistsError, SkillNotFoundError, ) from skill_mcp.models_crud import SkillCrudInput from skill_mcp.services.skill_service import SkillService from skill_mcp.services.template_service import TemplateRegistry class SkillCrud: """Unified tool for skill CRUD operations.""" @staticmethod def get_tool_definition() -> list[types.Tool]: """Get tool definition.""" return [ types.Tool( name="skill_crud", description="""Unified CRUD tool for skill management. IMPORTANT NOTES: - Skills are stored in ~/.skill-mcp/skills directory - All file paths in responses are relative to the skill directory (e.g., 'main.py', not full paths) - To execute scripts, use the 'run_skill_script' tool, NOT external bash/shell tools **Operations:** - **create**: Create a new skill with templates (basic, python, bash, nodejs) - **list**: List all skills with optional search (supports text and regex) - **search**: Search for skills by pattern (text or regex) - **get**: Get detailed information about a specific skill - **validate**: Validate skill structure and get diagnostics - **delete**: Delete a skill directory (requires confirm=true) - **list_templates**: List all available skill templates with descriptions **Examples:** ```json // List available templates {"operation": "list_templates"} // Create a Python skill {"operation": "create", "skill_name": "my-skill", "description": "My skill", "template": "python"} // List all skills {"operation": "list"} // Search skills by text {"operation": "search", "search": "weather"} // Search skills by regex pattern {"operation": "search", "search": "^api-"} // Get skill details {"operation": "get", "skill_name": "my-skill", "include_content": true} // Validate skill {"operation": "validate", "skill_name": "my-skill"} // Delete skill {"operation": "delete", "skill_name": "my-skill", "confirm": true} ```""", inputSchema=SkillCrudInput.model_json_schema(), ) ] @staticmethod async def skill_crud(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle skill CRUD operations.""" operation = input_data.operation try: if operation == "list": return await SkillCrud._handle_list(input_data) elif operation == "search": return await SkillCrud._handle_search(input_data) elif operation == "get": return await SkillCrud._handle_get(input_data) elif operation == "validate": return await SkillCrud._handle_validate(input_data) elif operation == "delete": return await SkillCrud._handle_delete(input_data) elif operation == "create": return await SkillCrud._handle_create(input_data) elif operation == "list_templates": return await SkillCrud._handle_list_templates(input_data) else: return [ types.TextContent( type="text", text=f"Unknown operation: {operation}. Valid operations: create, list, search, get, validate, delete, list_templates", ) ] except Exception as e: return [types.TextContent(type="text", text=f"Error: {str(e)}")] @staticmethod async def _handle_list(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle list operation.""" all_skills = SkillService.list_skills() # Apply search filter if provided skills = all_skills if input_data.search: import re skills = [ s for s in all_skills if ( input_data.search.lower() in s.name.lower() or input_data.search.lower() in (s.description or "").lower() or ( re.search(input_data.search, s.name, re.IGNORECASE) if input_data.search.startswith("^") or "*" in input_data.search else False ) ) ] if not skills: result = "No skills found in ~/.skill-mcp/skills" if input_data.search: result += f" matching '{input_data.search}'" else: result = f"Found {len(skills)} skill(s)" if input_data.search: result += f" matching '{input_data.search}'" result += ":\n\n" for skill in skills: status = "✓" if skill.has_skill_md else "✗" result += f"{status} {skill.name}\n" if skill.description: result += f" Description: {skill.description}\n" result += "\n" return [types.TextContent(type="text", text=result)] @staticmethod async def _handle_search(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle search operation.""" if not input_data.search: return [ types.TextContent( type="text", text="Error: search pattern is required for 'search' operation" ) ] all_skills = SkillService.list_skills() # Apply search filter import re skills = [ s for s in all_skills if ( input_data.search.lower() in s.name.lower() or input_data.search.lower() in (s.description or "").lower() or ( re.search(input_data.search, s.name, re.IGNORECASE) if input_data.search.startswith("^") or "*" in input_data.search else False ) ) ] if not skills: result = f"No skills found matching '{input_data.search}'" else: result = f"Found {len(skills)} skill(s) matching '{input_data.search}':\n\n" for skill in skills: status = "✓" if skill.has_skill_md else "✗" result += f"{status} {skill.name}\n" if skill.description: result += f" Description: {skill.description}\n" result += "\n" return [types.TextContent(type="text", text=result)] @staticmethod async def _handle_get(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle get operation.""" if not input_data.skill_name: return [ types.TextContent( type="text", text="Error: skill_name is required for 'get' operation" ) ] details = SkillService.get_skill_details(input_data.skill_name) result = f"Skill: {details.name}\n" result += f"Description: {details.description or 'N/A'}\n\n" # SKILL.md content if input_data.include_content and details.skill_md_content: result += "=== SKILL.md Content ===\n" result += details.skill_md_content result += "\n\n" # Files result += f"Files ({len(details.files)}):\n" for file in details.files: # Format modification time modified_str = "" if file.modified: from datetime import datetime modified_dt = datetime.fromtimestamp(file.modified) modified_str = f", modified: {modified_dt.strftime('%Y-%m-%d')}" # Use namespaced path format namespaced_path = f"{details.name}:{file.path}" result += f" - {namespaced_path} ({file.size} bytes{modified_str})" if file.is_executable: result += " [executable]" if file.has_uv_deps is not None: result += f" [uv deps: {'yes' if file.has_uv_deps else 'no'}]" result += "\n" # Scripts if details.scripts: result += f"\nScripts ({len(details.scripts)}):\n" for script in details.scripts: # Use namespaced path format namespaced_path = f"{details.name}:{script.path}" result += f" - {namespaced_path} ({script.type})" if script.has_uv_deps: result += " [has uv dependencies]" result += "\n" # Environment variables result += "\nEnvironment Variables:\n" if details.env_vars: for var in details.env_vars: result += f" - {var}\n" else: result += " (none)\n" result += f"\n.env file exists: {'Yes' if details.has_env_file else 'No'}\n" return [types.TextContent(type="text", text=result)] @staticmethod async def _handle_validate(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle validate operation.""" if not input_data.skill_name: return [ types.TextContent( type="text", text="Error: skill_name is required for 'validate' operation" ) ] # Simple validation: check if skill exists and has SKILL.md from skill_mcp.core.config import SKILL_METADATA_FILE, SKILLS_DIR skill_dir = SKILLS_DIR / input_data.skill_name if not skill_dir.exists(): raise SkillNotFoundError(f"Skill '{input_data.skill_name}' does not exist") errors: list[str] = [] warnings: list[str] = [] # Check for SKILL.md skill_md = skill_dir / SKILL_METADATA_FILE if not skill_md.exists(): errors.append("SKILL.md file is missing") else: # Try to parse YAML frontmatter try: from skill_mcp.utils.yaml_parser import ( get_skill_description, parse_yaml_frontmatter, ) content = skill_md.read_text() metadata = parse_yaml_frontmatter(content) if not metadata: warnings.append("SKILL.md has no YAML frontmatter") elif not get_skill_description(metadata): warnings.append("SKILL.md missing description in frontmatter") except Exception as e: errors.append(f"Invalid YAML frontmatter: {str(e)}") is_valid = len(errors) == 0 result = f"Validation for skill '{input_data.skill_name}':\n" result += f"Status: {'✓ Valid' if is_valid else '✗ Invalid'}\n\n" if errors: result += "Errors:\n" for error in errors: result += f" - {error}\n" if warnings: result += "\nWarnings:\n" for warning in warnings: result += f" - {warning}\n" if is_valid: result += "\nSkill is valid and ready to use." return [types.TextContent(type="text", text=result)] @staticmethod async def _handle_delete(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle delete operation.""" if not input_data.skill_name: return [ types.TextContent( type="text", text="Error: skill_name is required for 'delete' operation" ) ] if not input_data.confirm: return [ types.TextContent( type="text", text=f"Error: confirm=true is required to delete skill '{input_data.skill_name}'", ) ] # Delete skill directory import shutil from skill_mcp.core.config import SKILLS_DIR skill_dir = SKILLS_DIR / input_data.skill_name if not skill_dir.exists(): raise SkillNotFoundError(f"Skill '{input_data.skill_name}' does not exist") shutil.rmtree(skill_dir) return [ types.TextContent( type="text", text=f"Successfully deleted skill '{input_data.skill_name}'" ) ] @staticmethod async def _handle_list_templates( input_data: SkillCrudInput, ) -> list[types.TextContent]: """Handle list_templates operation.""" templates = TemplateRegistry.list_templates() result = f"Available templates ({len(templates)}):\n\n" for name, spec in templates.items(): result += f"**{name}**\n" result += f" Description: {spec.description}\n" result += f" Files: {', '.join(spec.files)}\n\n" result += "Use template name in 'create' operation:\n" result += ' {"operation": "create", "skill_name": "my-skill", "template": "python"}' return [types.TextContent(type="text", text=result)] @staticmethod async def _handle_create(input_data: SkillCrudInput) -> list[types.TextContent]: """Handle create operation.""" if not input_data.skill_name: return [ types.TextContent( type="text", text="Error: skill_name is required for 'create' operation" ) ] # Validate template template = input_data.template or "basic" try: TemplateRegistry.validate_template(template) except InvalidTemplateError as e: return [types.TextContent(type="text", text=f"Error: {str(e)}")] # Create skill directory with SKILL.md from skill_mcp.core.config import SKILL_METADATA_FILE, SKILLS_DIR skill_dir = SKILLS_DIR / input_data.skill_name if skill_dir.exists(): raise SkillAlreadyExistsError(f"Skill '{input_data.skill_name}' already exists") skill_dir.mkdir(parents=True) # Create SKILL.md with YAML frontmatter description = input_data.description or f"{input_data.skill_name} skill" skill_md_content = f"""--- name: {input_data.skill_name} description: {description} --- # {input_data.skill_name} {description} """ skill_md_path = skill_dir / SKILL_METADATA_FILE skill_md_path.write_text(skill_md_content) files_created = ["SKILL.md"] # Add template-specific files if input_data.template == "python": script_path = skill_dir / "main.py" script_path.write_text( """#!/usr/bin/env python3 '''Main script for {skill_name}.''' def main(): print("Hello from {skill_name}!") if __name__ == "__main__": main() """.format(skill_name=input_data.skill_name) ) files_created.append("main.py") elif input_data.template == "bash": script_path = skill_dir / "main.sh" script_path.write_text( """#!/usr/bin/env bash # Main script for {skill_name} echo "Hello from {skill_name}!" """.format(skill_name=input_data.skill_name) ) script_path.chmod(0o755) files_created.append("main.sh") elif input_data.template == "nodejs": script_path = skill_dir / "main.js" script_path.write_text( """#!/usr/bin/env node // Main script for {skill_name} console.log("Hello from {skill_name}!"); """.format(skill_name=input_data.skill_name) ) files_created.append("main.js") # Create package.json package_json_path = skill_dir / "package.json" package_json_content = """{{ "name": "{skill_name}", "version": "1.0.0", "description": "{description}", "main": "main.js", "scripts": {{ "start": "node main.js" }}, "dependencies": {{}} }} """.format(skill_name=input_data.skill_name, description=description) package_json_path.write_text(package_json_content) files_created.append("package.json") return [ types.TextContent( type="text", text=f"Successfully created skill '{input_data.skill_name}' with {len(files_created)} files:\n" + "\n".join(f" - {f}" for f in files_created), ) ]