Read a web page

Send a URL and get the page back as clean Markdown, ready to hand to a model. No HTML parsing and no scraping code. It runs on the same key as your chat requests and is billed to the same account. It is a paid ability.

Reach for it when you want to ground a model on a page you know: an article, a doc or a changelog. To find pages you do not know yet, use Search the web.

From your agent or your code#

Add the Flux tools server once, and your agent gets a web_fetch tool that takes a url. From your own code, send the URL to POST /v1/fetch.

claude mcp add --transport http flux-tools https://api.fluxrouter.ai/mcp-tools/ \
  --header "Authorization: Bearer sk-YOUR-FLUX-KEY"

[mcp_servers.flux-tools]
url = "https://api.fluxrouter.ai/mcp-tools/"
bearer_token_env_var = "FLUX_API_KEY"

{
  "mcpServers": {
    "flux-tools": {
      "url": "https://api.fluxrouter.ai/mcp-tools/",
      "headers": {
        "Authorization": "Bearer sk-YOUR-FLUX-KEY"
      }
    }
  }
}

mcp_servers:
  flux-tools:
    url: "https://api.fluxrouter.ai/mcp-tools/"
    headers:
      Authorization: "Bearer sk-YOUR-FLUX-KEY"

{
  "mcpServers": {
    "flux-tools": {
      "type": "streamableHttp",
      "url": "https://api.fluxrouter.ai/mcp-tools/",
      "headers": {
        "Authorization": "Bearer sk-YOUR-FLUX-KEY"
      }
    }
  }
}

curl https://api.fluxrouter.ai/v1/fetch \
  -H "Authorization: Bearer sk-YOUR-FLUX-KEY" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://example.com" }'

import requests

resp = requests.post(
    "https://api.fluxrouter.ai/v1/fetch",
    headers={"Authorization": "Bearer sk-YOUR-FLUX-KEY"},
    json={"url": "https://example.com"},
)
resp.raise_for_status()
print(resp.json()["markdown"])

const resp = await fetch('https://api.fluxrouter.ai/v1/fetch', {
  method: 'POST',
  headers: {
    Authorization: 'Bearer sk-YOUR-FLUX-KEY',
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({ url: 'https://example.com' }),
});
if (!resp.ok) throw new Error(await resp.text());

const page = await resp.json();
console.log(page.markdown);

A successful read returns:

{
  "markdown": "# Example Domain\n\nThis domain is for use in illustrative examples...",
  "format": "markdown",
  "url": "https://example.com"
}

From the Flux CLI, the same read is one command:

fluxrouter fetch https://example.com

The full field list and every error are in the fetch reference.

Pages that need a browser#

Some pages build themselves with JavaScript, so a plain read comes back empty or missing content. Set "render": true and Flux renders the page in a browser before it extracts the text.

curl https://api.fluxrouter.ai/v1/fetch \
  -H "Authorization: Bearer $FLUX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://app.example.com/dashboard", "render": true }'

fluxrouter fetch https://app.example.com/dashboard --render

A rendered read costs more than a plain one. render is an API option: the web_fetch tool always does a plain read.

Cache and billing#

• You pay per page read: priced per call. A rendered read is priced separately from a plain one.
• The same URL read again within 6 hours comes from the cache, and a cached read is not billed.
• Failed reads and blocked URLs are not billed.
• A read that takes longer than 25 seconds fails with a 502.

Charges land on the same bill as the rest of your usage. See How billing works.

On the Free plan a read answers 402 and nothing is charged. Upgrade in Billing.

What it cannot read#

• Private addresses. Private, loopback, link-local and cloud metadata addresses are refused with 400 blocked target. Only public http and https URLs work.
• Major social networks. Their pages are refused with 400 social_blocked.
• Pages behind a sign-in. Reads do not log in, so you get whatever a signed-out visitor sees.

The endpoint always returns Markdown. It is built to feed a model, not to mirror raw HTML.

If something goes wrong#

Every read error, with its fix, is on Read a page problems.