An MCP server that enables AI assistants to conduct structured, persona-driven conversations for interview preparation, personal reflection, and coaching. It supports loading rich persona profiles from a roles directory (e.g., Role-CIO.md, Role-Interview-Evaluator.md) to transform AI behavior and expertise across sessions. The server provides built-in timer management for session duration, start/stop control, and status checks, and utilizes pre-defined session templates defined in session_types.yaml with clear goals and outcomes. It supports adaptive flow, allowing sessions to respond to what emerges naturally in conversation, and offers a session evaluation capability to provide detailed feedback on performance and communication. Security considerations include safe file handling with path validation and error handling. The architecture includes a main MCP server (mcp-persona-sessions.py), timer utilities (timer.py), and configuration options in config.yaml (persona_path, session_types_file, default_persona_file, evaluator_persona_file). Users can add custom personas and session types, and integrate with MCP clients (e.g., Claude Desktop) via the provided configuration guidance. Use cases span job interview prep, presentation rehearsal, difficult conversations, personal development, and creative exploration.