npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

falsify-mcp

v0.2.3

Published

FALSIFY MCP server — deterministic auditing of mathematical assertions with replayable evidence (the verifier's verifier). Path-allowlisted, zero-dependency, bundled offline demo core. Windows & macOS/Linux.

Readme

falsify-mcp v0.2.3 — 数学断言自动审计插件(可下载使用版)

FALSIFY:数学断言的确定性审计器——LLM 只决定"检查什么",结论永远由确定性重算给出,每条击杀附带可回放证据链。验证器的验证器。

安装建议(v0.2.3 起):优先 npx -y falsify-mcp(装进用户缓存,无权限问题);macOS 上 npm i -g 请避免使用 sudo(若已 sudo 安装,v0.2.3 已将全部写入迁移到 ~/.cache/falsify-mcp/,不再依赖包目录可写)。

它能做什么

| 工具 | 说明 | 成本 | |---|---|---| | audit_paper | 审计一篇论文(LLM 提取 + 确定性裁决) | 需完整核心 + LLM_API_KEY(约 ¥0.3/篇) | | audit_claims_file | 审计已有断言 JSON | 免费、离线 | | verify_probe | 当场裁决单条断言 | 免费、离线 | | list_claim_kinds | 列出支持的断言类型 | 免费 |

三种裁决:killed(重算为假 + 反例证据)/ passed(重算为真)/ unverifiable(无确定性检查器覆盖,绝不猜测)。

Roadmap

  • 0.3.0 待核门控:击杀分两级——「待核」与「已核」;待核击杀由独立复核器重算反例后升级为已核(防假击杀的第二道闸)
  • 0.3.x 独立复核器:第二套实现 / 符号引擎重新推导反例
  • 0.4.0 击杀证据导出:一键导出复核报告(供论文 / 审稿使用)

关于击杀可信度:击杀结论 = 确定性重算 + 具体反例证据,任何击杀可回放复核;极端情形建议人工复核。

快速开始(3 步)

macOS / Linux

# 1. 解压并安装
unzip falsify-mcp-0.2.3.zip
cd falsify-mcp-0.2.3
bash install.sh        # 检查 node、选核心、生成 config.json、自检

# 2. 直接命令行试跑(stdin 协议)
echo '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}' | node src/server.js

# 3. 接入客户端(Chatbox / Claude Desktop 等)

Windows(PowerShell 5.1+)

# 1. 解压 falsify-mcp-0.2.2.zip
# 2. 右键 install.ps1 → 使用 PowerShell 运行(被策略阻止时:右键文件 → 属性 → 解除锁定,
#    或先执行 Set-ExecutionPolicy -Scope Process Bypass)
#    向导会完成: node 检查 / 核心选择 / 生成 config.json / 自检
# 3. 双击 falsify-mcp.bat 随时自检;或直接接入客户端

接入 Chatbox(Windows):设置 → MCP → 添加服务器 → 类型 stdio:

  • command: node
  • args: C:\路径\falsify-mcp-0.2.3\src\server.js
  • env: FALSIFY_HOME(安装向导已写入 config.json)、LLM_API_KEY(可选)

若你的 MCP 客户端支持把 .bat 当 command 用:command: C:\路径\falsify-mcp.bat server 亦可。 注意:Windows 下安装向导生成 config.json 使用无 BOM UTF-8;若手写 config.json,请勿带 BOM(Node 的 JSON.parse 会拒绝)。

接入 Chatbox

设置 → MCP → 添加服务器 → 类型 stdio:

  • command: node
  • args: <插件绝对路径>/src/server.js
  • env: FALSIFY_HOME(安装向导已写入 config.json,可留空)、LLM_API_KEY(可选)

接入 Claude Desktop

claude_desktop_config.json:

{
  "mcpServers": {
    "falsify": {
      "command": "node",
      "args": ["/绝对路径/falsify-mcp-0.2.1/src/server.js"],
      "env": {
        "FALSIFY_HOME": "/绝对路径/你的FALSIFY核心",
        "LLM_API_KEY": "sk-...",
        "LLM_BASE_URL": "https://api.deepseek.com/v1/chat/completions",
        "LLM_MODEL": "deepseek-chat"
      }
    }
  }
}

核心选择(两种模式)

  1. 自带演示核心 core_example/(默认,开箱即用) 覆盖离线确定性工具:verify_probe / audit_claims_file / list_claim_kinds。 无需外部依赖,无需 API Key。适合体验与集成测试。

  2. 完整核心(指向你自己的 FALSIFY 代码根目录) audit_paper 需要 m1/llm_extractor.js 与 LLM_API_KEY。 在 install.sh 中选择 1,或手动编辑 config.json:

    {
      "FALSIFY_HOME": "/绝对路径/到/FALSIFY代码根目录",
      "allowed_dirs": ["/绝对路径/到/论文目录"]
    }

配置优先级

FALSIFY_HOME 环境变量 > config.json 中的 FALSIFY_HOME > 自带 core_example/。 allowed_dirs(config.json)可额外放行论文/断言文件目录。

安全(v0.2.0+)

  • 路径白名单:只能读取 FALSIFY_HOME 与 allowed_dirs 内文件;realpath 归一化,../ 与符号链接无法绕过
  • 类型校验:audit_paper 仅 .tex,audit_claims_file 仅 .json
  • paper_id 清洗:仅允许 [A-Za-z0-9_-],防路径穿越
  • 建议:只在可信会话中挂载;allowed_dirs 尽量小;不要把凭据放进 FALSIFY_HOME

已知边界(诚实声明)

  • 覆盖基于断言类型(taxonomy):类型外的断言标记 unverifiable,不猜测
  • V2.2-01 构造绑定:枚举探针必须用论文自身定义,参数不匹配会产生假击杀——击杀请对照原文复核
  • 裁决对同一断言集是确定性的;提取出的断言集合随 LLM 批次可能不同(严格可回放请用 temperature 0 + 归档提取快照)

开发

node src/server.js          # 启动(stdio)
npm pack                    # 打成 npm tarball(可选)

零运行时依赖(纯 Node 内置模块),Node ≥ 18,Windows / macOS / Linux。

License

MIT(仅插件壳;FALSIFY 审计引擎与商业授权另行约定)