Context.dev for your agent, through one MCP connection
Give this to your agent
set up looot for Context.dev jobs - https://looot.ai/mcp/context-dev/skill.mdConnect in 30 seconds
Pick the client you use and add looot once. Context.dev is then one of the tools behind it. Each client takes one step, then your agent does the rest.
01Claude Code
Run this in your terminal, then sign in from the browser when it asks.
Terminal commandclaude mcp add looot https://api.looot.ai/mcp --transport http
02Cursor
Create .cursor/mcp.json with this server, then sign in from the browser.
mcp.json{"mcpServers":{"looot":{"url":"https://api.looot.ai/mcp"}}}
03ChatGPT
Where your plan lets you add a connector, add this URL and sign in.
Connector URLhttps://api.looot.ai/mcp
Then tell your agent, in any client:
Agent setup line
set up https://looot.ai/skill.mdWhat to ask
Prompts for Context.dev
Paste one into your agent after you connect looot.
Brand data
Use looot to get the logo, colours and description of stripe.com with Context.dev.
Page to markdown
Use looot to scrape https://example.com/pricing with Context.dev and return clean markdown.
Screenshot
Use looot to take a screenshot of https://example.com with Context.dev.
Jobs
What your agent can do with Context.dev
Prices are the cheapest live endpoint for each job, read from the looot catalog.
Compare
Using Context.dev directly versus through looot
Only the looot side is filled in. Check the provider's docs for the direct side.
| Context.dev directly | Through looot | |
|---|---|---|
| Account | Gap. Check the provider's docs | Yes. One looot sign-in. You do not set up a separate Context.dev account. |
| Price | Gap. Check the provider's docs | Yes. Per call, shown before the run. |
| Finding the endpoint | Gap. Check the provider's docs | Yes. Your agent searches the looot catalog for the task. |
| Other providers | Gap. Check the provider's docs | Yes. The same connection reaches every provider in the catalog. |
FAQ
Context.dev MCP, answered
Context.dev publishes its own MCP server. looot is not that server. looot is one MCP connection that reaches Context.dev and every other provider in the catalog, billed per call. Use Context.dev's server when Context.dev is the only provider you need.
Add the looot MCP connection to Claude Code once, using the setup at https://looot.ai/skill.md. Then ask Claude Code to use looot for the job. Context.dev is one of the providers behind looot; your agent calls its endpoints through the looot MCP connection.
Add the looot MCP connection in Cursor's MCP settings, then ask the agent to use looot. Context.dev is one of the providers behind looot; your agent calls its endpoints through the looot MCP connection.
It works in any client that can add an MCP connection. Where your ChatGPT plan lets you add one, add looot and ask it to use looot. Context.dev is one of the providers behind looot; your agent calls its endpoints through the looot MCP connection.
The jobs on this page: read brand data for a domain, convert pages to markdown, crawl a site to markdown, search the web, take a screenshot, find company news. The page marks any job whose endpoints are not in the live catalog.
You pay per call. The price of each endpoint is shown before you run it, so your agent can check it first and pick the cheapest one that fits.
You connect to looot, not to Context.dev. looot makes the provider call for you and bills each call at the price shown before the run.
It searches the looot catalog for the task, reads the endpoint's parameters and price, and then calls it. You do not have to read Context.dev's docs first.
Providers that often go with Context.dev are Firecrawl, Jina, Steel, The Companies API, Exa. All of them sit behind the same looot MCP connection.
For pages to markdown, Firecrawl, Jina and Steel run behind the same looot connection. For company data from a domain, The Companies API. See /mcp/firecrawl, /mcp/jina and /mcp/thecompaniesapi, or ask your agent to compare prices in the looot catalog before it runs.
It reads public pages and public brand data. It cannot read pages behind a login you hold, and a brand with a thin website returns thin data. Crawls and searches are priced per result, so cap the page count your agent asks for.
For agents
- provider
- Context.dev
- role
- Web and brand data API
- job
- Read brand data for a domain: covered
- job
- Convert pages to markdown: covered
- job
- Crawl a site to markdown: covered
- job
- Search the web: covered
- job
- Take a screenshot: covered
- job
- Find company news: covered
- pair with
- firecrawl jina steel thecompaniesapi exa
- setup
- https://looot.ai/mcp/context-dev/skill.md
- llms.txt
- https://looot.ai/llms.txt
- updated
- 2026-10-02
markdown /mcp/context-dev.md
Give your agent Context.dev and every other provider.
Start your workspace, top up, and paste one prompt into the agent you already use.
