mcp -neurolora-p

Neurolorap is an MCP server that provides tools for code analysis and documentation generation. It exposes two main MCP tools: a code collection tool that can gather code from an entire project or specific paths and output Markdown with syntax highlighting, and a project structure reporter tool that analyzes a project and generates a detailed report. You can run the server in normal mode, which serves via the MCP protocol for programmatic access, or in developer mode (JSON-RPC terminal) for interactive tool usage. To interact with the tools, you can either run the server and call the tools programmatically through the MCP interface, or enter developer mode to execute commands like collect and report directly from a terminal session.

To use the tools, you can leverage the code_collector and project_structure_reporter integrations. For example, to collect code from the entire project, you would invoke the code_collector with input set to "." and provide a title. To generate a project structure report, you can call the project_structure_reporter with an output filename such as PROJECT_STRUCTURE_REPORT.md. The developer mode exposes commands like help, list_tools, collect <path>, report [path], and exit, which let you explore available tools and run them against specified paths.

Prerequisites:

• Python (compatible version as required by the project, typically Python 3.8+)
• Optional: UV for a smoother installation workflow if you choose to follow the UV-based setup in related projects

Installation options:

Option 1: Install via pip (recommended for Python users)

pip install mcp-server-neurolorap

Option 2: Run via Python module (from source repository or installed package)

python -m mcp_server_neurolorap

Option 3: If using UV (for alternative installation flow), follow the project’s UV-based guidance:

uvx mcp-server-neurolorap

Post-installation:

• Start the server and connect via MCP protocol as documented in the project README.
• The server stores generated files under ~/.mcp-docs/<project-name>/ and creates a .neurolora symlink in your project root for easy access. You can customize ignore patterns by adding a .neuroloraignore file in the project root.

Development setup (optional):

python -m venv .venv
source .venv/bin/activate  # Unix
# or
.venv\Scripts\activate  # Windows

pip install -e ".[dev]"

Run the server:

# Normal mode
python -m mcp_server_neurolorap

# Developer mode (JSON-RPC terminal interface)
python -m mcp_server_neurolorap --dev

Notes and tips:

• The server exposes two main MCP tools: code_collector and project_structure_reporter. Use the developer mode or MCP calls to interact with these tools.
• Default storage path for generated files is ~/.mcp-docs/<project-name>/; a .neurolora symlink is created in the project root to this directory for quick access.
• To customize behavior, add a .neuroloraignore file in the project root. If absent, a default ignore configuration is created.
• The server can be started in normal mode or developer mode. In developer mode, you get a JSON-RPC interface for direct tool invocation.
• Ensure you have prerequisites installed (Python and optional UV). If you plan to use UV-based installation, follow the uvx installation path mentioned in the project README.
• When using MCP tools, you can specify inputs as strings or lists to target multiple paths and provide titles for the output artifacts.