Browserbase
Give an agent a browser session.
Browser automation can involve remote browser sessions, navigation, screenshots or interactive actions. A service may document only some of these capabilities. Start with the published candidates below and review the precise action, transport and access requirements before connecting a browser to an agent.
Alphabetical order. These are published metadata records, not a performance ranking. Pricing, connection requirements and evidence status remain visible on each profile.
Give an agent a browser session.
Control a browser from an agent.
Control browser sessions through an API-key-authenticated MCP endpoint.
Create narrated web-app walkthrough videos through MCP.
Scrape, screenshot or render pages using wallet payments.
Check whether the operation runs in a remote session or a local browser, whether sessions persist, and where any credentials or downloaded data will live. A hosted website is not itself proof of a remote browser API.
Read the provider documentation for screenshots, navigation, form interaction or other actions your task requires. Do not infer a capability from the word browser alone.
Review the host’s approval model before allowing submissions, purchases or account changes. Catalogue discovery grants no permission for a later consequential browser action.
Check the exact listing and provider connection guide. Some operations require a provider key or host-managed login. ToolSkipper does not inspect or create those connections when you browse a listing.
No. Screenshot capture must be documented for the particular endpoint or tool. A general browser description or evidence from a different transport is insufficient.
Read source links and check dates before relying on a service. ToolSkipper discovery never connects accounts or authorizes a payment.
Connect your agent →