Browser

A real browser: read a page after its JavaScript has run, or photograph it

Calling it

Metered methods answer an unauthenticated call with 402 and an x402 challenge — pay it and the same request succeeds, with no account at all. A token works too.

curl -H "Authorization: Bearer $MU_TOKEN" \
  https://micro.mu/api/v1/browser/read

The same method over MCP, which is what an agent speaks: browser_read. The answer is identical.

How every call behaves →

Methods

3 creditsGET /api/v1/browser/read

Open a page in a real browser and read it after its JavaScript has run. Use it when web_fetch comes back empty or with only a nav bar, which is what a page that builds itself in the browser looks like to a plain fetch. web_fetch is free and is the right first try; this costs, because it runs a browser

Draws 3 credits per call, or accepts payment.

Argument	Type	Description
`url` *	string	The page to open
`wait`	string	Optional CSS selector to wait for before reading, for content that arrives late

curl \
  -H "Authorization: Bearer $MU_TOKEN" \
  "https://micro.mu/api/v1/browser/read?url=url"

url *wait

Costs 3 credits

4 creditsGET /api/v1/browser/shot

Photograph a page and get back a URL for the picture. What it looks like, rather than what it says — a chart, a layout, a page whose content is an image. Ask for full to capture past the first screenful

Draws 4 credits per call, or accepts payment.

Argument	Type	Description
`url` *	string	The page to photograph
`full`	boolean	Capture the whole page rather than one screenful
`wait`	string	Optional CSS selector to wait for before the shot

curl \
  -H "Authorization: Bearer $MU_TOKEN" \
  "https://micro.mu/api/v1/browser/shot?url=url"

url *fullwait

Costs 4 credits

The same methods are tools over MCP, and every service on this instance is in one reference. Same method, same answer, same price.