This server provides tools for AI agents to scrape and compress web content into token-efficient markdown (70–90% fewer tokens than raw HTML).
Compress a single URL (scraper_compress_url): Fetch and compress any web page into clean markdown, with configurable fetch mode, timeout, and retry attempts (0–10).
Crawl multiple pages (scraper_crawl_url): Starting from a seed URL, perform BFS link discovery and compress up to 100 pages, with control over max depth, include/exclude URL glob patterns, timeout, and fetch mode.
Flexible fetch strategies: Choose from fast (plain HTTP), stealth (TLS fingerprint impersonation for anti-bot sites), render (headless browser for JS-rendered SPAs), or auto (fast with stealth fallback on 403/challenge).
Local, free, and unlimited: All operations run locally (API keys stay on your machine), with no cost or usage tiers.
Converts web pages and crawled website content into token-efficient Markdown, providing clean, compressed text specifically optimized for AI agent context windows.
⚠️ DEPRECATED — folded into
@robot-resources/scraper. This MCP server now ships as a built-inscraper-mcpbin inside the main scraper package. Install one package, get the MCP. Auto-config vianpx robot-resources --for=cursorornpx robot-resources --for=claude-code. The tools (scraper_compress_url,scraper_crawl_url) and parameters below are unchanged.
MCP server for Scraper — context compression for AI agents.
Human Resources, but for your AI agents.
Robot Resources gives AI agents two superpowers:
Router — Routes each LLM call to the cheapest capable model. 60-90% cost savings across OpenAI, Anthropic, and Google.
Scraper — Compresses web pages to clean markdown. 70-80% fewer tokens per page.
Both run locally. Your API keys never leave your machine. Free, unlimited, no tiers.
npx robot-resourcesOne command sets up everything. Learn more at robotresources.ai
Related MCP server: Scraper MCP
This package gives AI agents two tools to compress web content into token-efficient markdown via the Model Context Protocol: single-page compression and multi-page BFS crawling.
npx @robot-resources/scraper-mcpOr install globally:
npm install -g @robot-resources/scraper-mcpAdd to your claude_desktop_config.json:
{
"mcpServers": {
"scraper": {
"command": "npx",
"args": ["-y", "@robot-resources/scraper-mcp"]
}
}
}scraper_compress_urlCompress a single web page into markdown with 70-90% fewer tokens.
Parameters:
Parameter | Type | Required | Default | Description |
| string | yes | — | URL to compress |
| string | no |
|
|
| number | no |
| Fetch timeout in milliseconds |
| number | no |
| Max retry attempts (0-10) |
Example prompt: "Compress https://docs.example.com/getting-started"
scraper_crawl_urlCrawl multiple pages from a starting URL using BFS link discovery.
Parameters:
Parameter | Type | Required | Default | Description |
| string | yes | — | Starting URL to crawl |
| number | no |
| Max pages to crawl (1-100) |
| number | no |
| Max link depth (0-5) |
| string | no |
|
|
| string[] | no | — | URL patterns to include (glob) |
| string[] | no | — | URL patterns to exclude (glob) |
| number | no |
| Per-page timeout in milliseconds |
Example prompt: "Crawl the docs at https://docs.example.com with max 20 pages"
Mode | How | Use when |
| Plain HTTP | Default sites, APIs, docs |
| TLS fingerprint impersonation | Anti-bot protected sites |
| Headless browser (Playwright) | JS-rendered SPAs |
| Fast → stealth fallback on 403/challenge | Unknown sites (default) |
Stealth requires impit and render requires playwright as peer dependencies of @robot-resources/scraper.
Node.js 18+
@robot-resources/scraper - Core compression library
@robot-resources/router-mcp - MCP server for LLM cost optimization
Robot Resources - Human Resources, but for your AI agents
MIT
Compress web content from a URL for reduced token usage. Returns markdown with 70-90% fewer tokens than raw HTML.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | URL to compress | |
| mode | No | Fetch mode: 'fast' (plain HTTP), 'stealth' (TLS fingerprint), 'render' (headless browser), 'auto' (fast with fallback). Default: 'auto' | |
| timeout | No | Fetch timeout in milliseconds (default: 10000) | |
| maxRetries | No | Max retry attempts (default: 3) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. It mentions the token reduction benefit but fails to disclose critical traits like rate limits, authentication needs, error handling, or what happens with invalid URLs. The description is minimal and does not compensate for the lack of annotations, leaving significant gaps in understanding the tool's behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is highly concise and front-loaded, consisting of two sentences that directly state the tool's function and outcome without any fluff. Every sentence earns its place by delivering essential information efficiently, making it easy to grasp the tool's purpose quickly.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity of a tool with 4 parameters, no annotations, and no output schema, the description is incomplete. It lacks details on behavioral aspects, error cases, and output format beyond 'markdown,' failing to provide sufficient context for safe and effective use. The description does not compensate for the absence of structured data, leaving key operational questions unanswered.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 100% description coverage, providing clear details for all parameters (url, mode, timeout, maxRetries). The description adds no additional parameter semantics beyond what the schema already explains, such as usage examples or implications of mode choices. Thus, it meets the baseline score of 3, as the schema adequately covers parameter information.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose with a specific verb ('compress') and resource ('web content from a URL'), and distinguishes it from its sibling 'scraper_crawl_url' by focusing on token reduction rather than crawling. It explicitly mentions the outcome: 'Returns markdown with 70-90% fewer tokens than raw HTML.'
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus its sibling 'scraper_crawl_url' or any alternatives. It lacks context about prerequisites, such as when compression is beneficial or when to choose specific modes, offering only a basic functional statement without comparative or exclusionary advice.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Crawl multiple pages from a starting URL using BFS link discovery. Returns compressed markdown for each page with 70-90% fewer tokens than raw HTML.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | Starting URL to crawl | |
| maxPages | No | Max pages to crawl (default: 10) | |
| maxDepth | No | Max link depth (default: 2) | |
| mode | No | Fetch mode: 'fast' (plain HTTP), 'stealth' (TLS fingerprint), 'render' (headless browser), 'auto' (fast with fallback). Default: 'auto' | |
| include | No | URL patterns to include (glob) | |
| exclude | No | URL patterns to exclude (glob) | |
| timeout | No | Per-page timeout in milliseconds (default: 10000) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden and does well by disclosing key behavioral traits: the BFS crawling algorithm, multi-page scope, and the 70-90% token reduction in output. It doesn't mention rate limits, authentication needs, or error handling, but provides substantial operational context.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences that efficiently convey the core functionality and key output characteristic. Every word earns its place with no redundancy or unnecessary elaboration. The description is front-loaded with the primary action and scope.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a 7-parameter tool with no annotations and no output schema, the description provides good operational context but lacks details about return format structure, error conditions, or performance characteristics. It mentions compressed markdown output but doesn't describe the data structure or pagination approach.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the baseline is 3. The description adds no parameter-specific information beyond what's already in the schema descriptions. It mentions 'starting URL' which is covered by the url parameter description, but doesn't elaborate on parameter interactions or usage patterns.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose with specific verbs ('crawl multiple pages', 'returns compressed markdown') and resources ('starting URL', 'BFS link discovery'). It distinguishes from the sibling tool scraper_compress_url by emphasizing multi-page crawling rather than single-page compression.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage context through 'crawl multiple pages' and 'starting URL', suggesting this is for web scraping tasks. However, it doesn't explicitly state when to use this tool versus the sibling scraper_compress_url or other alternatives, nor does it provide exclusion criteria.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
The two tools have clearly distinct purposes: scraper_compress_url handles single-page compression, while scraper_crawl_url performs multi-page crawling with BFS link discovery. There is no overlap or ambiguity between them.
Both tools follow a consistent verb_noun pattern (scraper_compress_url and scraper_crawl_url), using the same prefix and snake_case style throughout. The naming is predictable and uniform.
With only two tools, the server feels thin for a scraper domain, lacking essential operations like error handling, configuration, or content filtering. This minimal set may limit agent workflows and cause dead ends.
The tool surface is severely incomplete for web scraping; it misses basic CRUD-like operations such as updating crawl parameters, deleting cached data, or retrieving specific page elements. Agents will struggle with incomplete coverage.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Web scraping for AI agents. Converts URLs to clean, LLM-ready Markdown with anti-bot bypass.
Cloud scraping & crawling API for AI agents. Turn any URL into clean, LLM-ready markdown.
Web scraping to clean Markdown with JS rendering, multi-page crawl, structured extract, sitemaps.
Turn any URL into clean Markdown and structured data. Scrape, crawl, search and extract.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/robot-resources/scraper-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server