writestat

WriteStat MCP Server analyzes text readability and detects AI-like patterns. It exposes tools to assess readability metrics such as Flesch-Kincaid and SMOG, identify hard sentences, flag AI-style phrases, and optionally run ML-based AI-detection. Use it by installing the writestat-mcp package and running it through the uvx runner to expose the MCP server under the writestat-mcp name. Once running, you can feed your text to the tools via the MCP interface to get structured analyses, batch analyses, and before/after comparisons to guide rewrites and improvements.

Prerequisites:

• Python 3.10+ (recommended)
• pip (comes with Python)

Installation steps:

1. Install the MCP server package (optionally with ML support):

pip install writestat-mcp

# Optional ML detectors (~500MB for torch/transformers)
pip install writestat-mcp[ml]

2. Prepare NLTK data (as required by the package):

python -c "import nltk; nltk.download('punkt_tab')"

3. Run or expose via MCP runner (uvx):

# See server commands in the README; using uvx as shown in Claude setup
uvx writestat-mcp

4. Interact with the MCP server using the specified tooling once it is running.

Notes and tips:

• ML detection is optional and requires additional dependencies; install writestat-mcp[ml] to enable it.
• If you plan to integrate with Claude or other tooling, you can register the MCP under a local mcpServers entry (see mcp_config) and point clients to writestat-mcp.
• Ensure Python 3.10+ is active in your environment when running the server to avoid compatibility issues.
• If nltk data download fails in automated environments, consider pre-downloading or providing the data path to the runtime.
• The project focuses on readability metrics, hard-sentence detection, AI-detection patterns, and multi-text processing via batch_analyze.