Skip to main content
Glama

@copperline/rendex-mcp

npm version npm downloads License: MIT

用于 Rendex 的 MCP 服务器 —— 通过使用模型上下文协议 (Model Context Protocol) 的 AI 代理,截取任何网页的屏幕截图和 PDF。

快速开始

Claude Desktop / Cursor / Windsurf (npx)

添加到您的 MCP 客户端配置中:

{
  "mcpServers": {
    "rendex": {
      "command": "npx",
      "args": ["-y", "@copperline/rendex-mcp"],
      "env": {
        "RENDEX_API_KEY": "your-api-key"
      }
    }
  }
}

添加位置:

客户端

配置位置

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json (macOS)

Cursor

项目根目录下的 .cursor/mcp.json,或 Settings > MCP

Windsurf

Settings > MCP Servers

Claude Code (CLI)

在项目根目录添加一个 .mcp.json,内容与上述配置相同。然后重启 Claude Code。

重要提示:将 .mcp.json 添加到您的 .gitignore 中 —— 它包含您的 API 密钥。

远程 (无需安装)

直接连接 —— 无需安装(仅限 Claude Desktop):

{
  "mcpServers": {
    "rendex": {
      "url": "https://mcp.rendex.dev/mcp",
      "headers": {
        "Authorization": "Bearer your-api-key"
      }
    }
  }
}

Related MCP server: Screenshot API

工具

rendex_screenshot

截取任何网页或原始 HTML 的屏幕截图或 PDF。

"Take a screenshot of https://example.com"
"Capture the full page of https://news.ycombinator.com in dark mode"
"Generate a PDF of https://github.com with A4 page size"
"Capture https://amazon.de as seen from Germany"
"Render this HTML invoice as a PDF"

参数:

参数

类型

默认值

描述

url

string

必填*

要截取的网页 URL。与 html 互斥。

html

string

—

要渲染的原始 HTML。与 url 互斥。

format

"png"

"jpeg"

"webp"

"pdf"

"png"

输出格式

fullPage

boolean

false

截取整个可滚动页面

darkMode

boolean

false

模拟深色配色方案

width

number

1280

视口宽度 (320-3840)

height

number

800

视口高度 (240-2160)

quality

number

80

图像质量 1-100 (仅限 JPEG/WebP,默认 80)

delay

number

0

截取前等待的毫秒数

blockAds

boolean

true

屏蔽广告和追踪器

blockResourceTypes

string[]

—

屏蔽资源类型:font, image, media, stylesheet

deviceScaleFactor

number

2

设备像素比 (1-3)。默认 2× Retina

timeout

number

30

等待页面加载的最长秒数 (5-60)

waitUntil

string

"networkidle2"

页面就绪状态:load, domcontentloaded, networkidle0, networkidle2

waitForSelector

string

—

截取前等待的 CSS 选择器

bestAttempt

boolean

true

超时时返回部分渲染结果而不是失败

selector

string

—

要截取的元素的 CSS 选择器(代替全页截取)

css

string

—

截取前注入的自定义 CSS (最大 50KB)

js

string

—

截取前执行的自定义 JavaScript (最大 50KB)

cookies

array

—

用于身份验证截取的 Cookie (最大 50)

headers

object

—

页面请求的自定义 HTTP 标头

userAgent

string

—

覆盖浏览器用户代理字符串

pdfFormat

string

—

PDF 页面尺寸:A4, Letter, Legal, Tabloid, A3

pdfLandscape

boolean

—

PDF 横向模式

pdfPrintBackground

boolean

true

在 PDF 中打印背景

pdfScale

number

1

PDF 缩放比例 (0.1-2)

pdfMargin

object

—

PDF 页边距:以 CSS 值表示的 {top, right, bottom, left}

geo

string

—

用于地理位置定位截取的 ISO 国家代码 (Pro/Enterprise)

geoCity

string

—

地理位置定位的城市 (需要 geo)

geoState

string

—

地理位置定位的州/省 (需要 geo)

async

boolean

—

异步处理 (返回任务 ID)

webhookUrl

string

—

异步截取完成时接收回调的 URL

cacheTtl

number

—

缓存结果的秒数 (3600-2592000)

身份验证

在 rendex.dev 获取您的 API 密钥。

在您的 MCP 客户端配置中设置 RENDEX_API_KEY 环境变量。

定价

计划

每月调用次数

速率

免费版

500

10/分钟

入门版

10,000

60/分钟

专业版

100,000

300/分钟

企业版

自定义

1,000/分钟

许可证

MIT — Copperline Labs LLC

Available Tools

11 tools
rendex_extractAInspect

Extract clean reader-mode content from any webpage as Markdown, JSON, or HTML. Runs the same Chromium render pass as a screenshot, so it captures content after JavaScript runs — handles SPAs that fetch-only readers miss. Strips nav, ads, and boilerplate, returning the article body plus title, byline, and excerpt. Great for feeding page content to an LLM, summarization, or RAG ingestion.

Inputs: url (required), extractFormat, waitUntil, timeout, +4 more.

rendex_screenshotAInspect

Capture a screenshot or PDF of any webpage, raw HTML, or Markdown. Supports full-page capture, dark mode, ad blocking, custom viewports, CSS/JS injection, cookie/header injection, PDF output, HTML and Markdown rendering, and progressive fallback for heavy sites. Returns partial renders on timeout by default (bestAttempt mode).

Inputs: url, html, markdown, format, +36 more.

watch_createBInspect

Create a Rendex Watch — monitor a URL on a schedule and get notified when it changes (real-Chrome visual diff with a highlighted overlay, an extracted-text diff, or both). An active watch captures its baseline immediately. Returns the created watch as JSON.

Inputs: url (required), name, intervalMinutes, diffMode, +6 more.

watch_deleteAInspect

Delete a watch and its run history. Irreversible.

Inputs: id (required).

watch_getAInspect

Fetch one watch by ID, including its current baseline image URL and status. Returns JSON.

Inputs: id (required).

watch_listAInspect

List your watches (newest first), optionally filtered by status and paged. Returns { items, nextCursor }.

Inputs: status, cursor, limit.

watch_runAInspect

Run an immediate check now (charges 1 credit). Returns the queued run; poll watch_runs for the result or receive a watch.changed webhook.

Inputs: id (required).

watch_runsAInspect

Read a watch's run history (newest first), paged. Each run includes changed, diffScore, and signed before/after/overlay image URLs. Returns { items, nextCursor }.

Inputs: id (required), cursor, limit.

watch_testAInspect

Dry-run a watch config BEFORE creating it — render the proposed config once and report what was captured + whether the page is reachable (and the text a text-watch would compare). Creates no watch, no baseline, no diff. Use this to validate a selector/scope/identity first. Returns JSON.

Inputs: url (required), name, intervalMinutes, diffMode, +6 more.

watch_updateAInspect

Update a watch in place — pause/resume (paused), re-point (url), change schedule/diff/notify settings, or turn a channel off (webhookUrl/notifyEmail = null). Only the fields you send change; renderParams is deep-merged over the existing config. A scope change (url/selector/fullPage/size/device) re-baselines on the next check. Returns the updated watch as JSON.

Inputs: id (required), url, name, intervalMinutes, +7 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4/5.0

    Scored across 11 tools

    Disambiguation4/5

    The two tool families are clearly separated (rendering vs. watch management), and each watch operation has a distinct role. However, rendex_screenshot and rendex_render_link share nearly the same rendering pipeline and options, and watch_run vs. watch_runs differ by only one letter, so a couple of tools could still be misselected without careful reading.

    Naming Consistency4/5

    All tool names use snake_case and a recognizable family prefix: rendex_* for rendering and watch_* for watch management. The imperative verb pattern is mostly consistent within the watch family, though watch_runs is a noun rather than a verb and rendex_screenshot reads as a noun, creating minor inconsistency.

    Tool Count5/5

    At 11 tools, the surface is well-scoped and reasonable for the server's dual purpose: three rendering/capture tools and eight watch lifecycle tools. Each tool has a clear job, and none feel redundant or unnecessary.

    Completeness5/5

    The watch family is fully fleshed out with create, read, list, update, delete, test, run, and history operations. The rendering family covers capture, extraction, and hosted-link generation, which covers the apparent domain without obvious dead ends.

    Maintenance

    ActivityNo data
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI agents to capture any public URL as PNG, JPEG, or PDF via REST API or MCP tools, including screenshot capture, page description, and PDF rendering.
      12 npm
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Hosted, SSRF-safe, cached screenshots and Open Graph images for AI agents - no headless Chrome to run.
      36 npm
      MIT
    • A
      license
      A
      quality
      C
      maintenance
      Screenshot, visual-diff, and AI page-analysis API for AI agents. Capture any URL as PNG, JPEG, WebP, PDF, or HTML, diff two versions of a page to catch visual regressions, and get an AI summary of what a page contains.
      3
      58 npm
      1
      MIT