claude-mpm 3.9.9__py3-none-any.whl → 4.0.3__py3-none-any.whl

claude_mpm/services/agents/deployment/agent_format_converter.py

@@ -0,0 +1,453 @@
+"""Agent Format Converter Service
+
+This service handles format conversion between different agent file formats,
+particularly YAML to Markdown conversion and format migration utilities.
+
+Extracted from AgentDeploymentService as part of the refactoring to improve
+maintainability and testability.
+"""
+
+import logging
+import re
+from datetime import datetime
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+from claude_mpm.core.logging_config import get_logger
+
+
+class AgentFormatConverter:
+    """Service for converting agent files between different formats.
+
+    This service handles:
+    - YAML to Markdown conversion with YAML frontmatter
+    - Format migration and backward compatibility
+    - Field extraction from various formats
+    - Content structure transformation
+    """
+
+    def __init__(self):
+        """Initialize the agent format converter."""
+        self.logger = get_logger(__name__)
+
+    def convert_yaml_to_md(self, target_dir: Path) -> Dict[str, Any]:
+        """
+        Convert existing YAML agent files to MD format with YAML frontmatter.
+
+        This method handles backward compatibility by finding existing .yaml
+        agent files and converting them to .md format expected by Claude Code.
+
+        Args:
+            target_dir: Directory containing agent files to convert
+
+        Returns:
+            Dictionary with conversion results
+        """
+        results = {"converted": [], "errors": [], "skipped": []}
+
+        if not target_dir.exists():
+            self.logger.debug(f"Target directory does not exist: {target_dir}")
+            return results
+
+        try:
+            # Find YAML files that need conversion
+            yaml_files = list(target_dir.glob("*.yaml"))
+
+            for yaml_file in yaml_files:
+                try:
+                    # Check if corresponding MD file already exists
+                    md_file = yaml_file.with_suffix(".md")
+
+                    if md_file.exists():
+                        # Check modification times
+                        yaml_mtime = yaml_file.stat().st_mtime
+                        md_mtime = md_file.stat().st_mtime
+
+                        if md_mtime >= yaml_mtime:
+                            # MD file is newer or same age, skip conversion
+                            results["skipped"].append(yaml_file.name)
+                            continue
+                        else:
+                            # MD file is older, proceed with conversion
+                            self.logger.info(
+                                f"MD file {md_file.name} is older than YAML, converting..."
+                            )
+
+                    # Read YAML content
+                    yaml_content = yaml_file.read_text()
+
+                    # Convert to MD format
+                    md_content = self.convert_yaml_content_to_md(
+                        yaml_content, yaml_file.stem
+                    )
+
+                    # Write MD file
+                    md_file.write_text(md_content)
+
+                    # Remove original YAML file
+                    yaml_file.unlink()
+
+                    results["converted"].append(
+                        {"from": yaml_file.name, "to": md_file.name}
+                    )
+
+                    self.logger.info(f"Converted {yaml_file.name} to {md_file.name}")
+
+                except Exception as e:
+                    error_msg = f"Failed to convert {yaml_file.name}: {e}"
+                    results["errors"].append(error_msg)
+                    self.logger.error(error_msg)
+
+        except Exception as e:
+            error_msg = f"YAML to MD conversion failed: {e}"
+            self.logger.error(error_msg)
+            results["errors"].append(error_msg)
+
+        return results
+
+    def convert_yaml_content_to_md(self, yaml_content: str, agent_name: str) -> str:
+        """
+        Convert YAML agent content to MD format with YAML frontmatter.
+
+        Args:
+            yaml_content: Original YAML content
+            agent_name: Name of the agent
+
+        Returns:
+            Converted Markdown content with YAML frontmatter
+        """
+        # Extract fields from YAML content
+        name = self.extract_yaml_field(yaml_content, "name") or agent_name
+        description = (
+            self.extract_yaml_field(yaml_content, "description")
+            or f"{agent_name.title()} agent for specialized tasks"
+        )
+        version = self.extract_yaml_field(yaml_content, "version") or "1.0.0"
+        tools_line = (
+            self.extract_yaml_field(yaml_content, "tools")
+            or "Read,Write,Edit,Grep,Glob,LS"
+        )
+
+        # Convert tools string to list format if needed
+        if isinstance(tools_line, str):
+            if tools_line.startswith("[") and tools_line.endswith("]"):
+                # Already in list format
+                tools_list = tools_line
+            else:
+                # Convert comma-separated to list
+                tools = [tool.strip() for tool in tools_line.split(",")]
+                tools_list = str(tools).replace("'", '"')
+        else:
+            tools_list = '["Read", "Write", "Edit"]'
+
+        # Extract additional fields
+        model = self.extract_yaml_field(yaml_content, "model") or "sonnet"
+        author = (
+            self.extract_yaml_field(yaml_content, "author")
+            or "claude-mpm@anthropic.com"
+        )
+
+        # Extract instructions from YAML content
+        instructions = self._extract_instructions_from_yaml(yaml_content, agent_name)
+
+        # Build new YAML frontmatter
+        current_time = datetime.now().isoformat() + "Z"
+
+        new_frontmatter = f"""---
+name: {name}
+description: "{description}"
+version: "{version}"
+author: "{author}"
+created: "{current_time}"
+updated: "{current_time}"
+tags: ["{agent_name}", "mpm-framework"]
+tools: {tools_list}
+model: "{model}"
+---
+
+"""
+
+        return new_frontmatter + instructions
+
+    def extract_yaml_field(self, yaml_content: str, field_name: str) -> Optional[str]:
+        """
+        Extract a field value from YAML content.
+
+        Args:
+            yaml_content: YAML content string
+            field_name: Name of the field to extract
+
+        Returns:
+            Field value or None if not found
+        """
+        # Try to match multi-line field first (with | or >)
+        multiline_pattern = rf"^{field_name}:\s*[|>]\s*\n((?:[ \t]+.+\n?)*)"
+        multiline_match = re.search(multiline_pattern, yaml_content, re.MULTILINE)
+        if multiline_match:
+            # Extract indented content and remove common indentation
+            lines = multiline_match.group(1).split("\n")
+            # Remove empty lines at the end
+            while lines and not lines[-1].strip():
+                lines.pop()
+
+            if lines:
+                # Find minimum indentation (excluding empty lines)
+                min_indent = float("inf")
+                for line in lines:
+                    if line.strip():  # Skip empty lines
+                        indent = len(line) - len(line.lstrip())
+                        min_indent = min(min_indent, indent)
+
+                # Remove common indentation
+                if min_indent != float("inf"):
+                    dedented_lines = []
+                    for line in lines:
+                        if line.strip():  # Non-empty line
+                            dedented_lines.append(line[min_indent:])
+                        else:  # Empty line
+                            dedented_lines.append("")
+                    return "\n".join(dedented_lines).strip()
+
+        # Try to match field with various quote styles
+        patterns = [
+            rf'^{field_name}:\s*"([^"]*)"',  # Double quotes
+            rf"^{field_name}:\s*'([^']*)'",  # Single quotes
+            rf"^{field_name}:\s*([^\n\r]+)",  # No quotes
+        ]
+
+        for pattern in patterns:
+            match = re.search(pattern, yaml_content, re.MULTILINE)
+            if match:
+                value = match.group(1).strip()
+                if value:
+                    return value
+
+        return None
+
+    def convert_md_to_yaml(self, md_content: str) -> str:
+        """
+        Convert Markdown with YAML frontmatter back to pure YAML format.
+
+        Args:
+            md_content: Markdown content with YAML frontmatter
+
+        Returns:
+            Pure YAML content
+        """
+        if not md_content.strip().startswith("---"):
+            # No frontmatter, treat as plain YAML
+            return md_content
+
+        # Split frontmatter and markdown content
+        parts = md_content.split("---", 2)
+        if len(parts) < 3:
+            return md_content
+
+        frontmatter = parts[1].strip()
+        markdown_content = parts[2].strip()
+
+        # Add instructions field if there's markdown content
+        if markdown_content:
+            frontmatter += f"\ninstructions: |\n"
+            # Indent markdown content
+            for line in markdown_content.split("\n"):
+                frontmatter += f"  {line}\n"
+
+        return frontmatter
+
+    def detect_format(self, content: str) -> str:
+        """
+        Detect the format of agent content.
+
+        Args:
+            content: Agent file content
+
+        Returns:
+            Format type: 'markdown_yaml', 'yaml', 'json', or 'unknown'
+        """
+        content = content.strip()
+
+        if content.startswith("---") and "---" in content[3:]:
+            return "markdown_yaml"
+        elif content.startswith("{") and content.endswith("}"):
+            return "json"
+        elif ":" in content and not content.startswith("#"):
+            # Likely YAML if it has key-value pairs and doesn't start with markdown header
+            return "yaml"
+        else:
+            return "unknown"
+
+    def normalize_agent_content(
+        self, content: str, agent_name: str, target_format: str = "markdown_yaml"
+    ) -> str:
+        """
+        Normalize agent content to a specific format.
+
+        Args:
+            content: Original agent content
+            agent_name: Name of the agent
+            target_format: Target format ('markdown_yaml', 'yaml', 'json')
+
+        Returns:
+            Normalized content in target format
+        """
+        current_format = self.detect_format(content)
+
+        if current_format == target_format:
+            return content
+
+        # Convert to intermediate format first (markdown_yaml)
+        if current_format == "yaml":
+            intermediate = self.convert_yaml_content_to_md(content, agent_name)
+        elif current_format == "json":
+            intermediate = self._convert_json_to_md(content, agent_name)
+        else:
+            intermediate = content
+
+        # Convert to target format
+        if target_format == "yaml":
+            return self.convert_md_to_yaml(intermediate)
+        elif target_format == "json":
+            return self._convert_md_to_json(intermediate)
+        else:
+            return intermediate
+
+    def get_conversion_stats(self, target_dir: Path) -> Dict[str, Any]:
+        """
+        Get statistics about files that need conversion.
+
+        Args:
+            target_dir: Directory to analyze
+
+        Returns:
+            Dictionary with conversion statistics
+        """
+        stats = {
+            "total_files": 0,
+            "yaml_files": 0,
+            "md_files": 0,
+            "json_files": 0,
+            "needs_conversion": 0,
+            "formats": {},
+        }
+
+        if not target_dir.exists():
+            return stats
+
+        # Analyze all agent files
+        for file_path in target_dir.glob("*"):
+            if file_path.is_file() and file_path.suffix in [
+                ".yaml",
+                ".yml",
+                ".md",
+                ".json",
+            ]:
+                stats["total_files"] += 1
+
+                if file_path.suffix in [".yaml", ".yml"]:
+                    stats["yaml_files"] += 1
+                    # Check if corresponding .md file exists
+                    md_file = file_path.with_suffix(".md")
+                    if not md_file.exists():
+                        stats["needs_conversion"] += 1
+                elif file_path.suffix == ".md":
+                    stats["md_files"] += 1
+                elif file_path.suffix == ".json":
+                    stats["json_files"] += 1
+
+                # Detect format
+                try:
+                    content = file_path.read_text()
+                    format_type = self.detect_format(content)
+                    stats["formats"][format_type] = (
+                        stats["formats"].get(format_type, 0) + 1
+                    )
+                except Exception:
+                    stats["formats"]["unreadable"] = (
+                        stats["formats"].get("unreadable", 0) + 1
+                    )
+
+        return stats
+
+    def _extract_instructions_from_yaml(
+        self, yaml_content: str, agent_name: str
+    ) -> str:
+        """
+        Extract instructions from YAML content.
+
+        Args:
+            yaml_content: YAML content
+            agent_name: Agent name for default instructions
+
+        Returns:
+            Instructions text
+        """
+        # Try to extract instructions field
+        instructions = self.extract_yaml_field(yaml_content, "instructions")
+        if instructions:
+            return instructions
+
+        # Try to extract description as instructions
+        description = self.extract_yaml_field(yaml_content, "description")
+        if (
+            description and len(description) > 50
+        ):  # Long description might be instructions
+            return f"# {agent_name.title()} Agent\n\n{description}"
+
+        # Default instructions
+        return f"# {agent_name.title()} Agent\n\nThis agent provides specialized functionality for your tasks."
+
+    def _convert_json_to_md(self, json_content: str, agent_name: str) -> str:
+        """Convert JSON content to Markdown with YAML frontmatter."""
+        try:
+            import json
+
+            data = json.loads(json_content)
+
+            # Convert JSON data to YAML-like string for processing
+            yaml_lines = []
+            for key, value in data.items():
+                if isinstance(value, str):
+                    yaml_lines.append(f'{key}: "{value}"')
+                elif isinstance(value, list):
+                    yaml_lines.append(f"{key}: {json.dumps(value)}")
+                else:
+                    yaml_lines.append(f"{key}: {value}")
+
+            yaml_content = "\n".join(yaml_lines)
+            return self.convert_yaml_content_to_md(yaml_content, agent_name)
+
+        except Exception as e:
+            self.logger.error(f"Failed to convert JSON to MD: {e}")
+            return f"# {agent_name.title()} Agent\n\nConversion failed: {e}"
+
+    def _convert_md_to_json(self, md_content: str) -> str:
+        """Convert Markdown with YAML frontmatter to JSON."""
+        try:
+            import json
+
+            import yaml
+
+            if not md_content.strip().startswith("---"):
+                return json.dumps({"error": "No YAML frontmatter found"})
+
+            # Extract frontmatter
+            parts = md_content.split("---", 2)
+            if len(parts) < 3:
+                return json.dumps({"error": "Invalid frontmatter format"})
+
+            frontmatter = parts[1].strip()
+            markdown_content = parts[2].strip()
+
+            # Parse YAML frontmatter
+            data = yaml.safe_load(frontmatter)
+
+            # Add instructions from markdown content
+            if markdown_content:
+                data["instructions"] = markdown_content
+
+            return json.dumps(data, indent=2)
+
+        except Exception as e:
+            self.logger.error(f"Failed to convert MD to JSON: {e}")
+            return json.dumps({"error": str(e)})

claude_mpm/services/agents/deployment/agent_frontmatter_validator.py

@@ -0,0 +1,161 @@
+"""Agent frontmatter validation for deployment service.
+
+This module provides validation and repair functionality for agent frontmatter.
+Extracted from AgentDeploymentService to reduce complexity and improve maintainability.
+"""
+
+import logging
+from pathlib import Path
+from typing import Any, Dict
+
+
+class AgentFrontmatterValidator:
+    """Validates and repairs agent frontmatter."""
+
+    def __init__(self, logger: logging.Logger):
+        """Initialize the validator with a logger."""
+        self.logger = logger
+
+    def validate_and_repair_existing_agents(self, agents_dir: Path) -> Dict[str, Any]:
+        """
+        Validate and repair broken frontmatter in existing agent files.
+
+        This method scans existing .claude/agents/*.md files and validates their
+        frontmatter. If the frontmatter is broken or missing, it attempts to repair
+        it or marks the agent for replacement during deployment.
+
+        WHY: Ensures all existing agents have valid YAML frontmatter before deployment,
+        preventing runtime errors in Claude Code when loading agents.
+
+        Args:
+            agents_dir: Directory containing agent .md files
+
+        Returns:
+            Dictionary with validation results:
+            - repaired: List of agent names that were repaired
+            - replaced: List of agent names marked for replacement
+            - errors: List of validation errors
+        """
+        results = {"repaired": [], "replaced": [], "errors": []}
+
+        try:
+            # Import frontmatter validator
+            from claude_mpm.agents.frontmatter_validator import FrontmatterValidator
+
+            validator = FrontmatterValidator()
+
+            # Find existing agent files
+            agent_files = list(agents_dir.glob("*.md"))
+
+            if not agent_files:
+                self.logger.debug("No existing agent files to validate")
+                return results
+
+            self.logger.debug(
+                f"Validating frontmatter in {len(agent_files)} existing agents"
+            )
+
+            for agent_file in agent_files:
+                try:
+                    agent_name = agent_file.stem
+
+                    # Read agent file content
+                    content = agent_file.read_text()
+
+                    # Check if this is a system agent (authored by claude-mpm)
+                    # Only repair system agents, leave user agents alone
+                    if (
+                        "author: claude-mpm" not in content
+                        and "author: 'claude-mpm'" not in content
+                    ):
+                        self.logger.debug(
+                            f"Skipping validation for user agent: {agent_name}"
+                        )
+                        continue
+
+                    # Extract and validate frontmatter
+                    if not content.startswith("---"):
+                        # No frontmatter at all - mark for replacement
+                        self.logger.warning(
+                            f"Agent {agent_name} has no frontmatter, marking for replacement"
+                        )
+                        results["replaced"].append(agent_name)
+                        # Delete the file so it will be recreated
+                        agent_file.unlink()
+                        continue
+
+                    # Try to extract frontmatter
+                    try:
+                        end_marker = content.find("\n---\n", 4)
+                        if end_marker == -1:
+                            end_marker = content.find("\n---\r\n", 4)
+
+                        if end_marker == -1:
+                            # Broken frontmatter - mark for replacement
+                            self.logger.warning(
+                                f"Agent {agent_name} has broken frontmatter, marking for replacement"
+                            )
+                            results["replaced"].append(agent_name)
+                            # Delete the file so it will be recreated
+                            agent_file.unlink()
+                            continue
+
+                        # Validate frontmatter with the validator
+                        validation_result = validator.validate_file(agent_file)
+
+                        if not validation_result.is_valid:
+                            # Check if it can be corrected
+                            if validation_result.corrected_frontmatter:
+                                # Apply corrections
+                                correction_result = validator.correct_file(
+                                    agent_file, dry_run=False
+                                )
+                                if correction_result.corrections:
+                                    results["repaired"].append(agent_name)
+                                    self.logger.info(
+                                        f"Repaired frontmatter for agent {agent_name}"
+                                    )
+                                    for correction in correction_result.corrections:
+                                        self.logger.debug(f"  - {correction}")
+                            else:
+                                # Cannot be corrected - mark for replacement
+                                self.logger.warning(
+                                    f"Agent {agent_name} has invalid frontmatter that cannot be repaired, marking for replacement"
+                                )
+                                results["replaced"].append(agent_name)
+                                # Delete the file so it will be recreated
+                                agent_file.unlink()
+                        elif validation_result.warnings:
+                            # Has warnings but is valid
+                            for warning in validation_result.warnings:
+                                self.logger.debug(
+                                    f"Agent {agent_name} warning: {warning}"
+                                )
+
+                    except Exception as e:
+                        # Any error in parsing - mark for replacement
+                        self.logger.warning(
+                            f"Failed to parse frontmatter for {agent_name}: {e}, marking for replacement"
+                        )
+                        results["replaced"].append(agent_name)
+                        # Delete the file so it will be recreated
+                        try:
+                            agent_file.unlink()
+                        except Exception:
+                            pass
+
+                except Exception as e:
+                    error_msg = f"Failed to validate agent {agent_file.name}: {e}"
+                    self.logger.error(error_msg)
+                    results["errors"].append(error_msg)
+
+        except ImportError:
+            self.logger.warning(
+                "FrontmatterValidator not available, skipping validation"
+            )
+        except Exception as e:
+            error_msg = f"Agent validation failed: {e}"
+            self.logger.error(error_msg)
+            results["errors"].append(error_msg)
+
+        return results