Renderly MCP
# Renderly MCP
**Clean web screenshots & content for AI agents.** An [MCP](https://modelcontextprotocol.io) server that gives your AI agent two tools:
- **`screenshot`** — capture a clean PNG/JPEG of any web page. Cookie banners, ads, and chat widgets are auto-removed, so the image is clean for **vision models**.
- **`extract`** — get the clean main article content (Markdown / text / HTML) of a page, boilerplate stripped — ideal for **LLMs / RAG**.
Real Chromium rendering, so JavaScript-built pages work. Powered by [Renderly](https://renderly.rest).
## Setup
1. **Get a RapidAPI key** — subscribe to Renderly (free tier available):
- Screenshot & PDF API: https://rapidapi.com/xufei547/api/screenshot-pdf-api2
- Article Extraction API: https://rapidapi.com/xufei547/api/article-extraction-api
Both share one RapidAPI key (your `X-RapidAPI-Key`).
2. **Add the server to your MCP client.**
### Claude Desktop / Cursor
Add to your MCP config (`claude_desktop_config.json` or Cursor's `mcp.json`):
```json
{
"mcpServers": {
"renderly": {
"command": "uvx",
"args": ["--from", "git+https://github.com/xufei547/renderly-mcp", "renderly-mcp"],
"env": { "RAPIDAPI_KEY": "your-rapidapi-key" }
}
}
}
```
(`uvx` ships with [uv](https://docs.astral.sh/uv/) — it installs and runs the server straight from GitHub, no extra steps. Prefer pip? `pip install git+https://github.com/xufei547/renderly-mcp` then use `"command": "renderly-mcp"`.)
## Tools
| Tool | Args | Returns |
|------|------|---------|
| `screenshot` | `url`, `full_page`, `block_cookie_banners`, `format` (png/jpeg), `width`, `height` | image |
| `extract` | `url`, `format` (markdown/text/html), `include_tables` | clean content |
## Example prompts
- "Screenshot https://example.com and describe what's on the page."
- "Extract the article at <url> and summarize it."
## Configuration (env vars)
| Var | Default | Purpose |
|-----|---------|---------|
| `RAPIDAPI_KEY` | — | **Required.** Your RapidAPI key. |
| `RENDERLY_SCREENSHOT_HOST` | `screenshot-pdf-api2.p.rapidapi.com` | Screenshot API host (see your RapidAPI "Code Snippets"). |
| `RENDERLY_EXTRACT_HOST` | `article-extraction-api.p.rapidapi.com` | Extraction API host. |
| `RENDERLY_TIMEOUT` | `90` | Request timeout (seconds). |
## License
MIT
TDQS
Scored across 2 tools
Both tools target web pages but have clearly distinct purposes: 'extract' retrieves text content for LLMs, while 'screenshot' captures visual images for vision models. No overlap in functionality.
Tool names are single-word verbs ('extract', 'screenshot'), which is consistent and descriptive, though lacking a verb_noun pattern that could clarify the object (e.g., 'extract_page_content').
Two tools is minimal for a web page utility server. While they cover the core promise of content extraction and screenshots, additional tools for batch processing or download would strengthen the set.
The server offers both text and visual extraction for web pages, covering its stated purpose. Minor gaps exist, such as lacking batch processing or file-saving options, but agents can accomplish most tasks.