Novada-MCP Server

Novada MCP

23-tool MCP server: web search, extraction, scraping (129 platforms), proxy, and browser automation.

Works with Claude, Cursor, VS Code, Windsurf, and any MCP-compatible client.

Install

Claude Code:

claude mcp add novada -e NOVADA_API_KEY=your_key -- npx -y novada-mcp

Claude Desktop / Cursor / VS Code / Windsurf — add to your MCP config:

{
  "mcpServers": {
    "novada": {
      "command": "npx",
      "args": ["-y", "novada-mcp"],
      "env": { "NOVADA_API_KEY": "your_key" }
    }
  }
}

Get your key at novada.com.

Health Check

NOVADA_API_KEY=your_key npx novada-mcp health

Or call novada_health from any MCP client to verify which products are active on your key.

Tool Selection

Have a question or topic, no URL → novada_search
Have a URL to read              → novada_extract
Read an entire site (docs, wiki) → novada_crawl
Discover all URLs on a site     → novada_map
Multi-source research report    → novada_research
Structured data (Amazon, TikTok, LinkedIn, …) → novada_scrape
Verify a factual claim          → novada_verify
JS-heavy / bot-blocked raw HTML → novada_unblock
Login, click, fill forms        → novada_browser
Proxy credentials for your own requests → novada_proxy_*

Tools

Search & Content

`novada_search`

Web search via Google, Bing, DuckDuckGo, Yandex. Returns ranked titles + URLs + snippets.

Param	Required	Default	Notes
`query`	✓	—	Search query
`engine`		`google`	`google` `bing` `duckduckgo` `yandex`
`num`		`10`	1–20 results
`time_range`		—	`day` `week` `month` `year`
`country`		—	ISO country code
`include_domains`		—	Restrict to these domains (max 10)
`exclude_domains`		—	Exclude these domains (max 10)
`extract_options`		—	Auto-extract top-N URLs (saves a separate `novada_extract` call)

novada_search({query: "Claude MCP 2025", engine: "google", num: 5})
novada_search({query: "AI news", time_range: "week", include_domains: ["techcrunch.com"]})

`novada_extract`

Extract clean content from any URL. Batch up to 10 URLs. Auto-escalates static → JS render.

Param	Required	Default	Notes
`url`	✓	—	Single URL or array of up to 10 URLs
`render`		`auto`	`auto` `static` `render` `browser`
`format`		`markdown`	`markdown` `text` `html`
`fields`		—	Extract specific fields: `["price", "author", "rating"]`
`max_chars`		25000	Max content length (max 100000)

novada_extract({url: "https://docs.example.com/api"})
novada_extract({url: ["https://a.com", "https://b.com"], format: "markdown"})
novada_extract({url: "https://shop.com/product", fields: ["price", "availability"]})

`novada_crawl`

Crawl a site BFS/DFS up to 20 pages, extract content from each.

Param	Required	Default	Notes
`url`	✓	—	Starting URL
`max_pages`		`5`	1–20
`strategy`		`bfs`	`bfs` (breadth-first) or `dfs` (depth-first)
`select_paths`		—	Regex patterns to include: `["/docs/.*"]`
`exclude_paths`		—	Regex patterns to skip: `["/blog/.*"]`
`render`		`auto`	`auto` `static` `render`
`instructions`		—	Natural-language hint: `"only API reference pages"`

novada_crawl({url: "https://docs.example.com", max_pages: 10, select_paths: ["/docs/.*"]})
novada_crawl({url: "https://example.com", instructions: "only quickstart and API pages"})

`novada_map`

Discover all URLs on a site (sitemap.xml + BFS fallback). Fast. Doesn't read content.

Param	Required	Default	Notes
`url`	✓	—	Root URL
`limit`		`50`	Max URLs to return
`max_depth`		`2`	Link hops from root
`search`		—	Filter URLs matching this term

novada_map({url: "https://example.com", limit: 100})
novada_map({url: "https://docs.example.com", search: "api"})

`novada_research`

Multi-source research: generates 3–10 parallel queries, deduplicates, extracts top sources, returns cited report.

Param	Required	Default	Notes
`question`	✓	—	Research question (min 5 chars)
`depth`		`auto`	`quick`=3 queries, `deep`=5–6, `comprehensive`=8–10
`focus`		—	Scope hint: `"technical implementation"` `"recent news only"`

novada_research({question: "How do MCP servers work with Claude?", depth: "deep"})
novada_research({question: "Best proxy providers 2025", focus: "pricing and reliability"})

`novada_verify`

Verify a factual claim against live web sources. Returns verdict + confidence score.

Verdicts: supported / unsupported / contested / insufficient_data

Param	Required	Default	Notes
`claim`	✓	—	Claim to verify (min 10 chars)
`context`		—	`"as of 2024"` `"in the US"`

novada_verify({claim: "OpenAI released GPT-5 in 2025"})
novada_verify({claim: "The Eiffel Tower is 330m tall", context: "as of 2024"})

Scraping (13 Active Platforms)

`novada_scrape`

Structured data from 13 active platforms. Synchronous — returns results immediately.

Param	Required	Default	Notes
`platform`	✓	—	Exact domain: `amazon.com` `tiktok.com` `linkedin.com`
`operation`	✓	—	Operation ID (see table below)
`params`	✓	`{}`	Operation-specific params
`format`		`markdown`	`markdown` `json` `toon`
`limit`		`20`	Max records returned

Active platforms and operations:

Platform	Operation ID	Required Params
`amazon.com`	`amazon_product_keywords`	`{ keyword }`
`amazon.com`	`amazon_product_asin`	`{ asin }`
`amazon.com`	`amazon_product_url`	`{ url }`
`amazon.com`	`amazon_comment_url`	`{ url }`
`walmart.com`	`walmart_product_keywords`	`{ keyword }`
`walmart.com`	`walmart_product_url`	`{ url }`
`google.com`	`google_search`	`{ q }`
`google.com`	`google_shopping_keywords`	`{ keyword }`
`bing.com`	`bing_search`	`{ keyword }`
`duckduckgo.com`	`duckduckgo`	`{ keyword }`
`yandex.com`	`yandex`	`{ keyword }`
`x.com`	`twitter_profile_username`	`{ username }`
`x.com`	`twitter_post_posturl`	`{ url }`
`tiktok.com`	`tiktok_posts_url`	`{ url }`
`tiktok.com`	`tiktok_profiles_url`	`{ url }`
`instagram.com`	`ins_profiles_username`	`{ username }`
`instagram.com`	`ins_posts_profileurl`	`{ url }`
`instagram.com`	`ins_comment_posturl`	`{ url }`
`facebook.com`	`facebook_post_posts-url`	`{ url }`
`facebook.com`	`facebook_profile_profiles-url`	`{ url }`
`youtube.com`	`youtube_video_search_label`	`{ label }`
`youtube.com`	`youtube_video-url`	`{ url }`
`youtube.com`	`youtube_comment_id`	`{ video_id }`
`linkedin.com`	`linkedin_company_information_url`	`{ url }`
`linkedin.com`	`linkedin_job_listings_information_keyword`	`{ keyword }`
`github.com`	`github_repository_repo-url`	`{ url }`
`github.com`	`github_repository_search-url`	`{ url }`

NOT AVAILABLE (return error 11006 — use novada_extract instead): reddit.com, glassdoor.com, zillow.com, ebay.com, etsy.com, tripadvisor.com, and ~100 others.

For the full platform+operation list: call novada_scrape and read the novada://scraper-platforms resource.

novada_scrape({platform: "amazon.com", operation: "amazon_product_keywords", params: {keyword: "iphone 16"}})
novada_scrape({platform: "linkedin.com", operation: "linkedin_company_information_url", params: {url: "https://linkedin.com/company/openai"}})
novada_scrape({platform: "github.com", operation: "github_repository_repo-url", params: {url: "https://github.com/anthropics/anthropic-sdk-python"}})

Async Scraping (`novada_scraper_submit` / `novada_scraper_status` / `novada_scraper_result`)

For long-running or batch scrape jobs. Use when novada_scrape times out.

1. task_id = novada_scraper_submit({platform: "amazon.com", operation: "amazon_product_asin", params: {asin: "B09..."}})
2. Poll: novada_scraper_status({task_id})  ← every 5–10s until status = "complete"
3. novada_scraper_result({task_id, format: "json"})

`novada_unblock`

Raw HTML from bot-protected or JS-heavy pages when you need the DOM.

Param	Required	Default	Notes
`url`	✓	—	Target URL
`method`		`render`	`render` or `browser`
`country`		—	Route through this country
`wait_for`		—	CSS selector to wait for

novada_unblock({url: "https://example.com/protected", method: "render"})

Proxy

All proxy tools return connection credentials (not proxied content). Use these in your own HTTP client.

Tool	Best for
`novada_proxy_residential`	Anti-bot bypass, geo-targeting (100M+ IPs)
`novada_proxy_isp`	Social media, e-commerce (looks like real home user)
`novada_proxy_mobile`	Mobile-targeted content, app APIs
`novada_proxy_datacenter`	High-volume scraping of unprotected targets (fastest)
`novada_proxy_static`	Account management, login-dependent workflows (sticky IP)
`novada_proxy_dedicated`	Clean reputation, exclusive IP
`novada_proxy`	Generic (choose type via param)

Common params: country (ISO code), session_id (sticky session), format (url/env/curl).

novada_proxy_residential({country: "us", format: "curl"})
novada_proxy_static({country: "de", session_id: "acct42", format: "env"})

Browser Automation

Requires NOVADA_BROWSER_WS env var (Browser API WebSocket).

`novada_browser`

Interactive cloud browser — navigate, click, type, screenshot, eval JS.

novada_browser({
  actions: [
    {action: "navigate", url: "https://example.com"},
    {action: "click", selector: "#login"},
    {action: "type", selector: "input[name=email]", text: "[email protected]"},
    {action: "screenshot"}
  ]
})

`novada_browser_flow`

Declarative multi-step flows with scroll, wait, and conditional steps.

Health & Discovery

Tool	Use when
`novada_health`	A tool is failing — check which products are active on your key
`novada_health_all`	Full endpoint health + latency for all services in parallel
`novada_discover`	Need the full tool catalog with categories and availability

Error Reference

Code	Meaning	Fix
`11006`	Invalid operation ID, or Scraper API not activated	Verify op ID against table above. If correct, activate at dashboard.novada.com
`11008`	Unknown platform name	Use exact domain (`amazon.com`, not `amazon`)
`27202`	Task still pending	Normal — poll again in 5s
`27203`	Server-side task failure	Transient — retry once
`10001`	Missing required params	Check `params` object for the operation

Environment Variables

Variable	Required	Purpose
`NOVADA_API_KEY`	Yes	Scraper API key (search + 129 platforms + web unblocker)
`NOVADA_BROWSER_WS`	Optional	Browser API WebSocket → enables `novada_browser` tools
`NOVADA_PROXY_USER`	Optional	Proxy username → enables `novada_proxy_*` tools
`NOVADA_PROXY_PASS`	Optional	Proxy password
`NOVADA_PROXY_ENDPOINT`	Optional	Proxy host:port

Load Only What You Need

Set NOVADA_GROUPS to limit which tool groups are registered (reduces token overhead in tool-heavy contexts):

{ "NOVADA_GROUPS": "search,extract,research" }

Valid groups: search extract crawl map research scrape proxy verify unblock browser health discover scraper_submit scraper_status scraper_result proxy_residential proxy_isp proxy_datacenter proxy_mobile proxy_static proxy_dedicated browser_flow

Common Workflows

Find and read pages on a topic:

novada_search({query: "..."}) → novada_extract({url: [top URLs]})

Deep research on a question:

novada_research({question: "...", depth: "deep"})

Scrape Amazon products then read reviews:

novada_scrape({platform: "amazon.com", operation: "amazon_product_keywords", params: {keyword: "..."}})
→ novada_scrape({platform: "amazon.com", operation: "amazon_comment_url", params: {url: product_url}})

Crawl a docs site:

novada_map({url: "https://docs.example.com"})   ← find URLs
→ novada_crawl({url: "...", select_paths: ["/docs/.*"], max_pages: 20})

Get LinkedIn company info + job listings:

novada_scrape({platform: "linkedin.com", operation: "linkedin_company_information_url", params: {url: "..."}})
novada_scrape({platform: "linkedin.com", operation: "linkedin_job_listings_information_keyword", params: {keyword: "..."}})

CLI

NOVADA_API_KEY=your_key npx novada-mcp search "GPT-5 release" --engine google --num 5
NOVADA_API_KEY=your_key npx novada-mcp extract https://example.com --format markdown
NOVADA_API_KEY=your_key npx novada-mcp research "AI agent frameworks 2025" --depth deep
NOVADA_API_KEY=your_key npx novada-mcp scrape --platform amazon.com --operation amazon_product_keywords --keyword "iphone 16"
NOVADA_API_KEY=your_key npx novada-mcp health

Documentation

Novada MCP

Install

Health Check

Tool Selection

Tools

Search & Content

`novada_search`

`novada_extract`

`novada_crawl`

`novada_map`

`novada_research`

`novada_verify`

Scraping (13 Active Platforms)

`novada_scrape`

Async Scraping (`novada_scraper_submit` / `novada_scraper_status` / `novada_scraper_result`)

`novada_unblock`

Proxy

Browser Automation

`novada_browser`

`novada_browser_flow`

Health & Discovery

Error Reference

Environment Variables

Load Only What You Need

Common Workflows

CLI

Links

Novada-MCP Server

Documentation

Novada MCP

Install

Health Check

Tool Selection

Tools

Search & Content

novada_search

novada_extract

novada_crawl

novada_map

novada_research

novada_verify

Scraping (13 Active Platforms)

novada_scrape

Async Scraping (novada_scraper_submit / novada_scraper_status / novada_scraper_result)

novada_unblock

Proxy

Browser Automation

novada_browser

novada_browser_flow

Health & Discovery

Error Reference

Environment Variables

Load Only What You Need

Common Workflows

CLI

Links

`novada_search`

`novada_extract`

`novada_crawl`

`novada_map`

`novada_research`

`novada_verify`

`novada_scrape`

Async Scraping (`novada_scraper_submit` / `novada_scraper_status` / `novada_scraper_result`)

`novada_unblock`

`novada_browser`

`novada_browser_flow`