Open Source

Libraries & MCP Servers

Veldica's open-source prose tooling: five libraries for Python and TypeScript, and one MCP server. The libraries split prose into structured blocks and score it for readability and style. PublishReady gives agents 16 callable tools for the same jobs.

Libraries

Python Prose Tokenizer

Zero-dependency Python library for tokenizing prose and Markdown. It splits text into Markdown-aware blocks and then sentences, with word and sentence counts for RAG chunking and LLM preprocessing.

Prose Tokenizer

The TypeScript counterpart: a rule-based tokenizer for English and Markdown that returns segmented paragraphs and sentences as structured data for AI pipelines and editorial tools.

Prose Analyzer

Deterministic style metrics for prose — lexical variety and pacing, plus sensory density. It runs locally and sends nothing anywhere, which makes it usable inside AI audits and editorial pipelines.

Prose Linter

Enforces style contracts and audits drafts for AI-writing markers. Built for programmatic content governance and agent feedback loops.

Readability

Readability formulas for English text: Flesch-Kincaid, Gunning Fog, SMOG, and Coleman-Liau. Same input, same scores, every run.

MCP Servers

PublishReady MCP

A local-first MCP server with 16 agent tools for writing analysis. It also ships as a CLI and a TypeScript toolkit, covering AI prose audits and target compliance.

This page is the index for the set. New libraries and MCP servers are added here as they are published.