Pagerunner — The Browser Layer That Learns
Drive real Chrome profiles that stay logged in, learn site behavior, and anonymize PII while agents automate authenticated web flows.
Overview
Pagerunner MCP is a Build-phase MCP server that runs real Chrome profiles with persistent logins, site intelligence, and PII anonymization for agent-driven browsing.
What is this MCP server?
- Real Chrome profiles with sessions that persist across agent runs
- Site intelligence layer that learns behavior on repeat visits
- Built-in PII anonymization for safer automated browsing
- npm stdio package pagerunner v0.8.0 — The Browser Layer That Learns
- Already-logged-in flows for dashboards, admin tools, and partner portals
Community signal: 3 GitHub stars.
What problem does it solve?
Headless one-off browsers break on login walls and forget context, forcing you to re-authenticate and re-teach agents the same sites every session.
Who is it for?
Indie builders integrating agents with authenticated web apps, internal admin UIs, or repeat operational browser tasks.
Skip if: Static API-only backends or teams that forbid real browser profiles and stored sessions on developer machines.
What do I get? / Deliverables
After registering pagerunner over stdio, agents reuse logged-in Chrome profiles with learned site behavior and safer PII handling across build and test flows.
- Reusable logged-in Chrome profiles for agent-driven web tasks
- Site intelligence that improves repeat automation on the same domains
- PII-reduced browsing outputs safer for logs and shared agent threads
Recommended MCP Servers
Journey fit
Browser-layer automation is built and wired into the product or agent stack during integration work, not during initial market research. Persistent profiles and site intelligence are integration concerns for agent tooling that touches the live web.
How it compares
Stateful Chrome-profile browser MCP, not a historical text corpus or censorship intelligence server.
Common Questions / FAQ
Who is Pagerunner for?
Solo developers and agent builders who need persistent, logged-in Chrome automation with PII controls instead of fragile headless scripts.
When should I use Pagerunner?
While building agent integrations that must interact with real sites, especially behind authentication, during development and integration testing.
How do I add Pagerunner to my agent?
Install the pagerunner npm package, add stdio MCP config per repository docs, and scope agent prompts to sites you are allowed to automate.