MCP Persona Sessions

What mattjoyce/mcp-persona-sessions MCP server does

The mattjoyce/mcp-persona-sessions MCP server organizes guided conversations around two configurable elements: a persona and a session framework. Personas define the simulated participant’s identity, communication style, expertise, and goals. Frameworks describe the session purpose, prerequisites, expected duration, topics, and desired outcomes.

The included examples cover meeting preparation, mock interviews, personal reflection, presentation rehearsal, difficult workplace conversations, creative exploration, and decision-making. Example personas include a healthcare CIO, board members, team leaders, reflective companions, and interview evaluators. The README also describes support for creating custom personas and session types.

How it works

An MCP-compatible assistant selects or initializes a session using the available frameworks and personas. Before a session begins, the server can check whether its prerequisites are satisfied. During the conversation, timer tools can start a duration-limited session, report elapsed progress, and stop the timer when the session concludes.

After the conversation, get_session_feedback analyzes the session transcript. The documented feedback use case is evaluating performance and communication effectiveness. The session flow is intended to remain adaptive, allowing the dialogue to respond to information that emerges during the exchange while retaining the selected structure.

Setup and configuration

The mattjoyce/mcp-persona-sessions MCP server is documented for Python 3.8 or later and an MCP-compatible assistant such as Claude Desktop. The repository setup uses a local clone, installation of the fastmcp dependency, and a copied config.yaml file. The MCP client configuration runs the repository’s Python script and sets the repository directory as the working directory.

The main configuration file identifies the persona directory, the session framework YAML file, a fallback persona, and the evaluator persona. The default example uses roles/ for persona files and session_types.yaml for framework definitions. Custom personas are Markdown files with sections for identity, communication style, expertise, and goals. Custom frameworks are added to the YAML file with a name, persona reference, description, prerequisites, duration, topics, and goals.

Tools and capabilities

The mattjoyce/mcp-persona-sessions MCP server exposes tools for the full session lifecycle:

• Browse available frameworks with list_session_frameworks.
• Discover personas with list_session_personas and inspect one with get_persona_details.
• Validate prerequisites using assess_session_readiness.
• Start a structured conversation with initialize_session.
• Manage timing with start_timer, check_timer_status, and stop_timer.
• Request transcript-based evaluation with get_session_feedback.

These tools let a client inspect the available options before starting and keep timing and evaluation within the same MCP workflow.

Limitations and notes

The documented installation is repository-based rather than a published package command. It requires local persona and framework files, and the session behavior depends on how those files are configured. The README does not describe hosted deployment, authentication, remote APIs, or integrations with services outside the MCP client and local project configuration. The listed client example is Claude Desktop, although the project describes its target more broadly as an MCP-compatible AI assistant.