内容
# PublishReady: Professional Writing Control
[](https://github.com/veldica/publishready-mcp/actions/workflows/ci.yml)
[](https://opensource.org/licenses/MIT)
**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](packages/mcp)**: The Model Context Protocol (MCP) server implementation (`publishready-mcp`).
- **[@veldica/publishready-cli](packages/cli)**: The command-line tool for local analysis (`publishready`).
- **[@veldica/publishready-core](packages/core)**: The central orchestration engine.
- **[@veldica/publishready-schemas](packages/schemas)**: Unified Zod schemas and explicit interfaces.
### Underlying Libraries
- **[@veldica/prose-analyzer](https://www.npmjs.com/package/@veldica/prose-analyzer)**: Deterministic style signals (variety, density, repetition, narrative texture).
- **[@veldica/readability](https://www.npmjs.com/package/@veldica/readability)**: Consolidated library of all major readability formulas.
- **[@veldica/prose-tokenizer](https://www.npmjs.com/package/@veldica/prose-tokenizer)**: Standalone markdown-aware prose tokenization.
- **[@veldica/prose-linter](https://www.npmjs.com/package/@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:
```json
{
"mcpServers": {
"publishready": {
"command": "npx",
"args": ["-y", "@veldica/publishready-mcp"]
}
}
}
```
### Command Line
```bash
npx @veldica/publishready-cli analyze sample.txt
```
### Hosted MCP
For Smithery, VPS, or gateway deployments, run the server with Streamable HTTP:
```bash
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`, and `reduce_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](packages/mcp/README.md).
## Deterministic Philosophy
This package explicitly avoids **perplexity** and other model-dependent scores. We believe writing control should be:
1. **Explainable**: You should know exactly why a score changed.
2. **Reproducible**: The same text should always yield the same metrics.
3. **Practical**: A metric is only useful if it tells you what to change.
## Development
```bash
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](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
[](https://smithery.ai/servers/veldica/publishready)
