Pagerunner vs Playwright MCP
Playwright MCP brings Microsoft's testing infrastructure to AI agents. Pagerunner brings site intelligence and privacy. Both persist auth now — the question is whether your tool learns from usage or just saves cookies.
Feature comparison
| Feature | Pagerunner | Playwright MCP |
|---|---|---|
| Site intelligence | Adapters, selector tracking | None |
| PII anonymization | Local ONNX NER | Raw content |
| Cross-browser | Chrome only | Chrome, Firefox, WebKit |
| Accessibility snapshots | DOM-based | Token-efficient |
| Persistent profiles | Named, encrypted | --storage-state |
| Session checkpoints | Full state | Cookies/localStorage only |
| LLM-aware responses | Metadata hints | Raw accessibility trees |
| KV store | Built-in | None |
| Backing | Independent | Microsoft |
| Price | Free (Apache-2.0) | Free (Apache-2.0) |
Where Pagerunner wins
- Site intelligence. Adapters and selector tracking mean your agent gets smarter with every session, not just more state files.
- PII anonymization. Sensitive data is stripped locally before it ever reaches an LLM.
- Full session checkpoints. Restore complete page state including scroll position, not just cookies.
- KV store. Built-in key-value storage enables multi-agent coordination and workflow state.
- LLM-aware response metadata. Hints help agents make better decisions with less context.
Where Playwright MCP wins
- Cross-browser. Supports Firefox and WebKit in addition to Chrome.
- Accessibility snapshots. Token-efficient page representation that can reduce LLM costs.
When to use which
Use Pagerunner when
You need authenticated workflows that learn over time, privacy protection, and rich state management. Best for production agent workflows that repeat across sessions.
Use Playwright MCP when
You need cross-browser testing, accessibility snapshots for token efficiency, or you're already invested in the Playwright ecosystem.