Summary
A local-first writing QA layer that runs deterministic style and readability checks without calling external APIs. It exposes 16 MCP tools including audit_ai_sounding_prose for formulaic phrase detection, analyze_with_template for matching genre conventions, and revision levers like shorten_long_sentences backed by specific evidence. Built on standalone packages for tokenization, readability formulas, and prose linting. Reach for this when you want reproducible metrics and actionable revision guidance on AI drafts before publishing, whether that's blog posts, documentation, or fiction. Works entirely locally over stdio, so your content stays private while you get concrete numbers on sentence variety, lexical density, dialogue ratio, and readability scores.
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
PublishReady: Professional Writing Control
PublishReady is a deterministic writing analysis system designed to turn AI drafts into publish-ready writing. It serves as the final QA pass for AI-generated prose, providing local-first metrics, target compliance, and specific revision levers without sending text to remote services.
The PublishReady Packages
This project is structured as a professional, layered monorepo containing specialized packages:
Core Packages
- @veldica/publishready-mcp: The Model Context Protocol (MCP) server implementation (
publishready-mcp). - @veldica/publishready-cli: The command-line tool for local analysis (
publishready). - @veldica/publishready-core: The central orchestration engine.
- @veldica/publishready-schemas: Unified Zod schemas and explicit interfaces.
Underlying Libraries
- @veldica/prose-analyzer: Deterministic style signals (variety, density, repetition, narrative texture).
- @veldica/readability: Consolidated library of all major readability formulas.
- @veldica/prose-tokenizer: Standalone markdown-aware prose tokenization.
- @veldica/prose-linter: Target checks, revision levers, content integrity, and deterministic AI-sounding prose markers.
Installation
MCP Server (Recommended)
Add the server to your MCP client configuration:
{
"mcpServers": {
"publishready": {
"command": "npx",
"args": ["-y", "@veldica/publishready-mcp"]
}
}
}
Command Line
npx @veldica/publishready-cli analyze sample.txt
Hosted MCP
For Smithery, VPS, or gateway deployments, run the server with Streamable HTTP:
npx @veldica/publishready-mcp --transport=http --port=3000
The MCP endpoint is /mcp; the health endpoint is /health.
Key Features
- Template, Target, and Reference Modes: Compare writing against built-in templates, explicit numeric targets, reference text, or reusable reference profiles.
- Deterministic Metrics: Structural counts, sentence and paragraph distributions, lexical signals, scannability, fiction proxies, and readability formulas.
- Specific Revision Levers: Ranked, evidence-based suggestions such as
shorten_long_sentences,replace_difficult_words, andreduce_abstract_wording. - AI-Sounding Prose Audit: Deterministic marker inventory for formulaic, generic, or over-polished prose, including exact matches and tracked phrase counts.
- Fiction & Non-Fiction Support: Narrative metrics for dialogue, sensory density, abstract wording, and scene pacing.
- Explainable Interpretation: Target and metric interpretation that explains audience, use cases, style implications, and tradeoffs.
- Local-First & Private: Stdio-first, deterministic, no external API calls, and no LLM wrappers.
MCP Tool Surface
The MCP server exposes 16 specialized tools for analysis and control, including audit_ai_sounding_prose for deterministic AI-marker analysis. For a full list and documentation, see the MCP README.
Deterministic Philosophy
This package explicitly avoids perplexity and other model-dependent scores. We believe writing control should be:
- Explainable: You should know exactly why a score changed.
- Reproducible: The same text should always yield the same metrics.
- Practical: A metric is only useful if it tells you what to change.
Development
npm install
npm run build
npm run lint
npm run typecheck
npm test
Publishing Metadata
- npm package:
@veldica/publishready-mcp - MCP Registry name:
io.github.veldica/publishready - Product homepage:
https://veldica.com/publish-ready - Source repository:
https://github.com/veldica/publishready-mcp
Registry and Directory Metadata
PublishReady is prepared for MCP directory discovery through:
- GitHub repository topics
- npm package keywords
- Glama metadata via
glama.json - Official MCP Registry metadata via
mcpName
This MCP is designed for AI-assisted writing workflows where the model should improve clarity, structure, readability, and publish-readiness while preserving facts, terminology, intent, and author voice.
License
MIT
Featured
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →