Fetch public web pages live and get clean Markdown, text or HTML.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Your Context Retrieval API key as a bearer token: Bearer 1337_live_…
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Check the remaining credits, plan and rate limit for the connected Context Retrieval API account. Use before large batches of fetches.
This tool takes no arguments.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http contextretrieval https://contextretrieval.net/mcp \ --header 'Authorization: Bearer YOUR_TOKEN'
Headers
Authorization (required): Your Context Retrieval API key as a bearer token: Bearer 1337_live_…Check the remaining credits, plan and rate limit for the connected Context Retrieval API account. Use before large batches of fetches.
No arguments.
{
"type": "object",
"properties": {}
}Fetch a public web page live and return its main content as clean Markdown, plain text or HTML. Use this when you have a specific URL and need its current content. Works best on articles, docs, product and listing pages. Does not log in, submit forms or execute JavaScript (yet).
| Argument | Type | Description |
|---|---|---|
| url* | string | The public URL to fetch (http or https, at most 8,192 characters). |
| format | string | Output format. Prefer markdown. |
| country | string | Fetch the page as seen from this country, e.g. 'de', 'gb', 'us'. |
| verbosity | string | 'compact' = main content only, 'full' = whole page body. |
| max_characters | integer | Maximum characters of content to return (default 8000, smaller than the REST API's 100000 to save context). Raise it when the content is cut. |
{
"type": "object",
"$schema": "http://json-schema.org/draft-07/schema#",
"required": [
"url"
],
"properties": {
"url": {
"type": "string",
"format": "uri",
"maxLength": 8192,
"description": "The public URL to fetch (http or https, at most 8,192 characters)."
},
"format": {
"enum": [
"markdown",
"text",
"html"
],
"type": "string",
"default": "markdown",
"description": "Output format. Prefer markdown."
},
"country": {
"type": "string",
"default": "us",
"maxLength": 2,
"minLength": 2,
"description": "Fetch the page as seen from this country, e.g. 'de', 'gb', 'us'."
},
"verbosity": {
"enum": [
"compact",
"full"
],
"type": "string",
"default": "compact",
"description": "'compact' = main content only, 'full' = whole page body."
},
"max_characters": {
"type": "integer",
"default": 8000,
"maximum": 100000,
"minimum": 500,
"description": "Maximum characters of content to return (default 8000, smaller than the REST API's 100000 to save context). Raise it when the content is cut."
}
}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 11, 2026.