OnlineCyberTools mcp

OnlineCyberTools mcp

MCP server that lets AI agents use the Online Cyber Tools catalogue as a set of native MCP tools.

Category
访问服务器

README

onlinecybertools-mcp-server

MCP (Model Context Protocol) server that lets AI agents — Claude Code, Codex, Cursor, Continue, etc. — use the Online Cyber Tools catalogue as a set of native MCP tools.

What it exposes

  • One MCP tool per documented MCP-compatible /api/tools/{category}/{tool} operation. POST tools use their JSON request-body schema; compatible GET tools use OpenAPI query/path parameters. Schemas are taken straight from the site's OpenAPI 3.1 spec at /api/openapi.json, so agents get per-tool argument validation.
  • A search meta-tool that performs the same keyword search humans use, backed by GET /api/tools/search?q=....
  • A describe_tool meta-tool that fetches the long page guidance, source links, page URL, API endpoint, and SEO description from GET /api/mcp/tool-docs/{tool_id}. It accepts either a menu ID such as ping or an MCP tool name such as network_ping.
  • A report_bug meta-tool that files a structured bug report against POST /api/agent/bug-report (hard rate-limited).

Calls are proxied to the live HTTP API — no algorithm is re-implemented here. That guarantees agents see whatever the deployed site does. Generated tool descriptions stay compact: menu ID, OpenAPI summary, OpenAPI description, and a hint to call describe_tool for full guidance.

<!-- TOOLS:BEGIN (auto-generated by php bin/console mcp:tools-doc; do not edit by hand) -->

Tools

This server exposes 280 tools across 15 categories: Encoding/Decoding, Binary/Text Conversion, Cryptography & Hashing, Web Dev Utilities, Text Utilities, OSINT Tools, Networking Tools, Security Tools, SEO Tools, Linux Tools, Date & Time, Math & Calculators, Data Tools, File Tools, Reverse Engineering.

This section and TOOLS.md are auto-generated from the live menu on every release, so the count and list never drift. <!-- TOOLS:END -->

Quick start

The package is published on npm as onlinecybertools-mcp-server, so any MCP client can launch it with npx -y onlinecybertools-mcp-server — no clone, no global install. The config snippets below work as-is and expose the full tool catalogue by default.

Prefer a guided setup? Generate a ready-made Claude Code plugin or Codex config block from the website's interactive builder:

https://onlinecybertools.com/integrations/mcp-plugin-builder

The builder lets you pick a subset of tools and emits the matching OCTOOLS_TOOLS filter for you (see Configuration).

Configuration

Configure via environment variables. All are optional.

Variable Default Purpose
OCTOOLS_BASE_URL https://onlinecybertools.com Site to proxy requests to.
OCTOOLS_TOOLS (unset → all tools) Comma-separated menu IDs (base64_encode,sha256,hash) to restrict the exposed surface. Leave unset to expose every tool — the examples below omit it on purpose.
OCTOOLS_STREAM_BYTE_CAP 262144 (256 KiB) Max bytes accumulated from a streamed (x-mcp-compatible: stream-buffered) endpoint.
OCTOOLS_STREAM_TIME_CAP_MS 30000 (30 s) Max wall-clock time spent buffering a streamed endpoint.

When OCTOOLS_TOOLS is set, the server appends ?tools=... to the spec fetch so the site returns a pre-filtered spec; the client also enforces the filter as defense-in-depth.

Running

Inspector (manual smoke test)

npx @modelcontextprotocol/inspector npx -y onlinecybertools-mcp-server

Open the inspector URL, click List Tools — you should see search, describe_tool, report_bug, plus one entry per compatible Symfony API operation. With no OCTOOLS_TOOLS set, the full catalogue is listed.

To hack on the server locally instead, clone and run from source:

git clone https://github.com/Jambozx/onlinecybertools-mcp-server.git
cd onlinecybertools-mcp-server
npm install
npx @modelcontextprotocol/inspector node index.mjs

Claude Code

Add to ~/.claude.json (or your project's .mcp.json):

{
  "mcpServers": {
    "octools": {
      "command": "npx",
      "args": ["-y", "onlinecybertools-mcp-server"]
    }
  }
}

This exposes every tool. To restrict the surface, add an env block with OCTOOLS_TOOLS:

{
  "mcpServers": {
    "octools": {
      "command": "npx",
      "args": ["-y", "onlinecybertools-mcp-server"],
      "env": {
        "OCTOOLS_TOOLS": "base64_encode,sha256,hash"
      }
    }
  }
}

Codex

Add to ~/.codex/config.toml:

[mcp_servers.octools]
command = "npx"
args = ["-y", "onlinecybertools-mcp-server"]

To restrict the surface, add an env line with OCTOOLS_TOOLS:

[mcp_servers.octools]
command = "npx"
args = ["-y", "onlinecybertools-mcp-server"]
env = { OCTOOLS_TOOLS = "base64_encode,sha256,hash" }

Cursor / Continue / generic MCP client

Most clients accept the same command/args/env shape. Point them at this package via npx -y onlinecybertools-mcp-server.

Streaming endpoints

Endpoints tagged x-mcp-compatible: stream-buffered in the spec (currently traceroute and proxy-test streams) are read to completion and returned as a single JSON envelope of accumulated SSE events. GET stream endpoints send tool arguments as query parameters; POST streams send JSON bodies. Hard caps:

  • 256 KiB of buffered output (OCTOOLS_STREAM_BYTE_CAP)
  • 30 s of wall-clock time (OCTOOLS_STREAM_TIME_CAP_MS)

Whichever cap fires first, the response envelope contains { "truncated": true } so the agent knows the output is partial.

Endpoints tagged x-mcp-compatible: none (multipart file uploads, etc.) are skipped at registration — they will not appear in tools/list.

Limitations

  • Spec is fetched once at startup. If the site adds new endpoints, restart the server.
  • stdio transport only; no HTTP server (avoids needing auth in front of a privileged endpoint).
  • Published to npm as onlinecybertools-mcp-server (npx -y onlinecybertools-mcp-server). Installing straight from GitHub (npx -y github:Jambozx/onlinecybertools-mcp-server) still works for the bleeding edge.

License

MIT.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选