Connect Crawlkit once and call the web like tools
Use one MCP server for web search, screenshot capture, TikTok, Instagram, LinkedIn, App Store, and Play Store workflows. Same credits, same auth model, no extra scraping infrastructure.
Use Crawlkit MCP for search, screenshot, TikTok, Instagram, LinkedIn, and app-store pipelines
Crawlkit MCP gives LLM agents one stable interface for structured web data, rendered pages, public social intelligence, and app-store review workflows.
Search and screenshot workflows
Run web search and full-page capture tools from the same MCP connection your agent already uses.
Social intelligence tools
Call TikTok, Instagram, and LinkedIn endpoints as normalized MCP tool invocations instead of wiring multiple APIs.
Store and review pipelines
Collect App Store and Play Store detail or review data without building a separate scraping stack.
How to connect Crawlkit MCP in Cursor, Claude Code, ChatGPT, or custom agents
Setup is intentionally simple: one URL, one bearer key, and the same Crawlkit credits you already use for direct API calls.
Add the Crawlkit MCP endpoint
Use Streamable HTTP transport and point your client to https://api.crawlkit.sh/mcp.
Authenticate with your ck_ key
Set the Authorization header as Bearer ck_xxx inside your MCP client configuration.
Call tools from prompts or agents
Use search, screenshot, social, and store tools from one MCP server without changing billing or auth models.
1{2 "mcpServers": {3 "crawlkit": {4 "transport": "streamable_http",5 "url": "https://api.crawlkit.sh/mcp",6 "headers": {7 "Authorization": "Bearer ck_your_api_key"8 }9 }10 }11}Everything you can run through the Crawlkit MCP server
Switch between capability groups and plug Crawlkit tools directly into your agent workflows.
Web Workflows via MCP
Use Crawlkit MCP tools for search and screenshot pipelines.
- Search the web and process normalized results
- Capture full-page screenshots with browser rendering
Preview real MCP JSON responses before you ship
Switch between tools and inspect the normalized payloads your MCP client will actually receive from Crawlkit.
Use these previews to shape prompts, schemas, and downstream parsing logic.
1{2 "success": true,3 "data": {4 "id": "kat***hig***s-a******",5 "name": "Katie H*****s",6 "first_name": "Katie",7 "last_name": "H*****s",8 "city": "Braintree, England, UK",9 "location": "Braintree",10 "country_code": "GB",11 "followers": 548,12 "connections": 500,13 "avatar": "https://static.licdn.com/...",14 "banner_image": "https://static.licdn.com/...",15 "linkedin_num_id": "2099250",16 "influencer": false,17 "default_avatar": true,18 "volunteer_experience": [{ ... }],19 "people_also_viewed": [{ ... }],20 "activity": [{ ... }],21 "creditsUsed": 2,22 "creditsRemaining": 984623 }24}Simple, predictable pricing for Crawlkit MCP usage
Usage-based credits with clear discount thresholds and no hidden infrastructure fees.
Credit tiers
MCP calls consume the same credits as direct Crawlkit REST requests, so billing stays predictable across clients and workflows.
- Starts at $0.96 per 1K credits
- No setup fee or MCP surcharge
- Lower effective rates at higher volume
| Usage (credits / month) | Price / 1K credits |
|---|---|
| 0 - 25K | $1.68 |
| 25K - 50K | $1.58 |
| 50K - 100K | $1.49 |
| 100K - 250K | $1.39 |
| 250K - 500K | $1.30 |
Reference packs: 25K=$42, 50K=$79, 100K=$149, 250K=$349, 500K=$649.
Frequently asked questions about Crawlkit MCP
Everything you need before using Crawlkit MCP in production.
Use https://api.crawlkit.sh/mcp with Streamable HTTP transport in your MCP client configuration.
Set Authorization header as Bearer ck_xxx inside your mcpServers.crawlkit.headers object.
MCP calls consume credits using the same endpoint pricing as direct REST API requests.
Search, screenshot, LinkedIn, Instagram, TikTok, Play Store, App Store, plus credits and API key validation. TikTok coverage includes search, shop, product, popular, and trending surfaces.
No. Credits never expire and remain available across all supported Crawlkit endpoints.
Keep the key only in your MCP client config, never commit it to public repos, and rotate it if exposure is suspected.