Etincel Nonfiction

What etincel-nonfiction MCP server does

The etincel-nonfiction MCP server adds nonfiction writing controls to Claude through the Model Context Protocol. Its stated focus is a trainable writing voice: the service is intended to help shape output around a selected or developed nonfiction style rather than treating every draft as generic text.

It also includes presets. Presets provide named or predefined writing configurations that can be applied to supported writing tasks. The available material does not describe how presets are created, whether they are stored between sessions, or which writing attributes they control.

A separate anti-AI-tells audit reviews writing for signals associated with AI-generated text. This makes the server relevant both during drafting and during revision, when an author wants to inspect a passage for those signals before using it.

Tools and capabilities

The documented capabilities are limited to three areas:

• Train or use a nonfiction writing voice.
• Apply writing presets.
• Run an anti-AI-tells audit.

These capabilities are described as working with Claude via MCP. The source material does not provide individual MCP tool names, input schemas, output formats, supported file types, or details about whether the audit returns explanations, scores, or rewritten text. Treat the capability list as functional guidance rather than a complete API reference.

The writing voice and presets address generation and style control. The audit addresses evaluation of existing writing. Together, they suggest a workflow in which a draft is produced or revised using a chosen voice and then checked for AI-associated phrasing. The available description does not establish that the server can publish, save, retrieve, or manage documents outside the Claude interaction.

Limitations and notes

The etincel-nonfiction MCP server has limited public detail in the supplied material. No README is available, so installation commands, transport options, authentication requirements, environment variables, licensing, and compatibility with specific MCP clients cannot be confirmed here.

The phrase “trainable” does not, by itself, establish whether voice training uses examples, instructions, persistent profile data, or session-only context. Similarly, “presets” does not identify a preset catalog or customization model. Developers should verify those behaviors on the linked source before building an automated workflow around them.

The anti-AI-tells audit should be treated as a review feature, not as evidence of authorship or a guarantee that text will avoid detection by every external system. No accuracy claims, detection methodology, or supported languages are provided.