> ## Documentation Index
> Fetch the complete documentation index at: https://docs.meshbrow.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Proxies

> Multi-provider proxy infrastructure in Meshbrow

# Proxies

Meshbrow routes each browser session through a proxy so your agents browse from realistic locations worldwide. The proxy system features automatic failover, quality scoring, and multi-provider aggregation.

## Architecture

```mermaid theme={null}
graph LR
    B[Browser] --> N[Network Namespace]
    N --> P[Proxy Router]
    P --> R[Residential Pool]
    P --> I[ISP Pool]
    P --> M[Mobile Pool]
    P --> D[Datacenter Pool]
    R --> W[Website]
    I --> W
    M --> W
    D --> W
```

## Quality Scoring

Every proxy is continuously scored on:

* **Success rate** — Percentage of requests that succeed
* **Speed** — Latency percentiles (p50, p95)
* **Detection rate** — How often the proxy triggers anti-bot
* **Ban rate** — How often the IP gets blocked
* **Uptime** — Availability over time

Low-scoring proxies are automatically deprioritized. Providers that fall below threshold are disabled with automatic recovery checks.

## Geo-Verification

Meshbrow verifies that proxies actually resolve to the requested geography:

```typescript theme={null}
// Request US proxy
const session = await client.sessions.create({
  proxy: { type: 'residential', country: 'US', city: 'New York' },
});

// Meshbrow verifies the assigned IP resolves to US before accepting
```

If geo-verification fails, the proxy is rejected and another is selected.

## Provider Failover

If a provider experiences degraded quality:

1. Quality score drops below threshold
2. Provider is disabled for new sessions
3. Recovery checks run periodically
4. Provider re-enabled when health recovers

This happens transparently — your sessions always get a working proxy.

## Cost Tracking

Proxy costs are tracked per-session:

* Bandwidth-based billing (residential, mobile)
* Per-IP billing (ISP, dedicated)
* Session-level cost attribution
* Provider-level spend analytics