Arachne MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MCP_HOST | No | Host untuk Streamable HTTP | 127.0.0.1 |
| MCP_PORT | No | Port untuk Streamable HTTP | 8000 |
| LLM_MODEL | No | Model LLM yang digunakan | |
| LLM_API_KEY | No | API key untuk LLM | |
| SEARXNG_URL | No | URL SearXNG server | http://localhost:8080 |
| LLM_BASE_URL | No | Base URL untuk LLM endpoint | |
| MCP_TRANSPORT | No | Transport MCP (stdio atau streamable-http) | stdio |
| ARACHNE_PROXY_URL | No | Proxy HTTP/HTTPS opsional | |
| ARACHNE_MAX_REQUESTS | No | Batas maksimum request per crawl | 200 |
| BRAVE_SEARCH_API_KEY | No | API key untuk Brave Search | |
| ARACHNE_RESPECT_ROBOTS | No | Mematuhi robots.txt | true |
| ARACHNE_MAX_TOTAL_CHARS | No | Batas output agregat | 250000 |
| ARACHNE_CACHE_TTL_SECONDS | No | TTL cache | 3600 |
| ARACHNE_ALLOW_BROWSER_EVAL | No | Mengizinkan arbitrary browser JS | false |
| ARACHNE_ALLOW_IGNORE_ROBOTS | No | Mengizinkan caller memilih respect_robots=false | false |
| ARACHNE_BROWSER_CONCURRENCY | No | Jumlah context Chromium bersamaan | 2 |
| ARACHNE_BLOCK_PRIVATE_NETWORKS | No | Proteksi SSRF | true |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| healthA | Return enabled engines, integrations, security policy, and parser availability. |
| crawl_urlB | Fetch and extract one URL using fast HTTP, browser rendering, or automatic escalation. Supports HTML, text, JSON, XML, PDF, DOCX, XLSX, CSV and TSV. Content returned to an agent is marked as untrusted and scanned for common prompt-injection text. |
| parse_urlC | Parse a remote document or non-HTML resource into LLM-ready text/Markdown. |
| crawl_siteB | Crawl a site with strict request/error/token budgets and per-host rate limiting. |
| map_siteC | Discover site URLs from recursive sitemaps plus optional shallow crawling. |
| extract_linksC | Extract normalized crawlable links from static or JavaScript-rendered pages. |
| check_robotsC | Inspect robots.txt permissions, crawl delay, request rate, and sitemap declarations. |
| browser_interactA | Operate a browser using safe declarative actions: click, fill, press, select, wait, scroll, and goto. JavaScript evaluate is disabled unless the server owner explicitly enables it. This tool does not bypass CAPTCHA, authentication, paywalls, or access controls. |
| batch_scrapeA | Scrape up to 500 URLs concurrently with a shared output budget. |
| search_webC | Search through a configured self-hosted SearXNG or Brave Search provider. |
| extract_structuredC | Extract JSON matching a caller-supplied schema through an optional OpenAI-compatible LLM. |
| crawl_diffB | Compare a page with its previous persistent snapshot and return a unified diff. |
| chunk_contentB | Split long content into stable, overlapping, hash-addressed chunks for RAG or agents. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 13 tools
Each tool serves a distinct purpose in the web scraping pipeline, from health checks to link extraction, crawling, parsing, site mapping, robots inspection, browser interaction, batch scraping, search, structured extraction, diffing, and chunking. No two tools have overlapping functionality.
All tool names follow a consistent verb_noun pattern in snake_case (e.g., extract_links, crawl_url, parse_url). The only exception is 'health', which is a noun but serves as a status probe and fits the convention of a single-word command.
13 tools is appropriate for a comprehensive web scraping and content extraction server. Each tool covers a necessary operation without redundancy, balancing breadth and focus.
The tool surface covers the full lifecycle of web data acquisition: discovery (map_site, search_web), fetching (crawl_url, crawl_site, batch_scrape), parsing (parse_url, extract_links), interaction (browser_interact), compliance (check_robots), processing (chunk_content, extract_structured), and monitoring (health, crawl_diff). No obvious gaps.