20 isolated identity slots
Each slot is a persistent identity slot: saved sessions, a consistent fingerprint, its own cookies, and logins that survive across agent runs.
MCP browser fleet
Give your AI agents a browser fleet
Shade is a hosted, MCP-native browser fleet: 20 isolated persistent identity slots over one connection, with saved sessions, leak checks, Web3 wallet monitoring, and a live console for human takeover.
Quickstart commandcopy
claude mcp add --transport http shade https://api.shade.atlas-ai.au/mcp --header "Authorization: Bearer KEY"Priced by persistent identity slot, not per-browser-hour. Your logins, fingerprint personality, and session state persist across agent runs.
Fleet primitives
Built for agents that return to the same work.
Ephemeral browser-hours are useful for one-off jobs. Shade is for agents that need a durable identity, remembered logins, a visible console, and a human who can step in without tearing down the run.
Save, load, and clone sessions
Agents can pause a logged-in workflow, resume it later, or clone a prepared browser state into another slot without starting from a blank browser.
Opt-in anti-detect personality
Fingerprint personality is opt-in and explicit. Shade does not promise bypass magic, CAPTCHA solving, or residential proxies.
IP, DNS, and WebRTC leak checks
Run a slot-level health check before giving an agent sensitive work. Surface mismatched egress, resolver, and WebRTC exposure early.
Web3 wallet monitoring
Watch wallet extension state, prompts, and connection surfaces from the fleet console so humans can approve or intervene.
Live fleet console with takeover
See every running slot in one console, step into a session when an agent gets stuck, and hand control back without breaking the MCP connection.
Synthetic action view
The live fleet console reconstructs what agents are doing from structured events, so supervisors can track many slots without streaming every page tree through the model.
Token-efficiency wedge
Watch the fleet without paying snapshot tax
Accessibility snapshots are useful when an agent needs page structure, but they are a heavy supervision primitive when a human only needs to watch progress across many slots. Shade's console reconstructs activity from structured events: navigation, clicks, form edits, downloads, wallet prompts, leak checks, and takeover state. That gives humans a synthetic action view without feeding every page tree back through the model.
Measured June 11, 2026 across 7 real pages (Wikipedia, GitHub, Stripe, NYTimes, MDN, HN, python.org): Shade's browser_snapshot is hard-capped at ~1,500 tokens regardless of page weight - 3-13x smaller than an uncapped accessibility tree (the same Playwright API playwright-mcp emits) and 100x+ smaller than raw HTML. On nytimes.com that is 1,281 tokens vs 553,782 for raw HTML. Reproducible: benchmarks/token_efficiency.py.
Comparison
Persistent slots are the product wedge.
Browserbase and Steel are strong browser cloud platforms. Playwright-MCP is a strong local MCP bridge. Shade is positioned around a remote, MCP-native fleet where identity slots are the primary unit.
| Capability | Shade | Browserbase | Steel | Playwright-MCP |
|---|---|---|---|---|
| Billing mental model | Persistent identity slot: saved sessions, consistent fingerprint, logins persist across runs | Usage/browser-hour pricing with session and context primitives | Usage/browser-hour pricing with session and profile primitives | Local browser process and host resources |
| Primary integration | MCP-native remote fleet over one connection | Browser platform with REST/SDK APIs and a Stagehand-based MCP integration | Sessions API and SDK-first browser cloud | Local MCP server launched per client |
| Identity continuity | Named slots designed to survive agent runs | Contexts can reuse cookies, authentication, and application data across sessions | Profiles and auth-context reuse can carry browser state across sessions | Depends on local profiles and client setup |
| Fleet visibility | One console for up to 20 slots plus human takeover | Session live view, recordings, replays, logs, and inspector tooling | Live and past session embeds plus agent trace surfaces | Local MCP introspection over the browser it launches |
| Token profile | Synthetic action view from structured events; benchmark coming soon | Browser/session telemetry depends on the chosen Browserbase integration | Session views and traces depend on the chosen Steel integration | Accessibility-tree snapshots are the core MCP interaction surface |
| Abuse-sensitive features | Opt-in personality, leak checks, no CAPTCHA solving, no residential proxies | Agent identity, verified browser, proxy, and CAPTCHA capabilities documented by Browserbase | Stealth, proxy, and CAPTCHA capabilities documented by Steel | No hosted anti-detect service |
Pricing teaser
Buy durable identities, not disposable browser time.
Starter
2 persistent identity slots
For one builder wiring Claude or Codex into a hosted browser that remembers logins.
$29/mo
- 2 saved browser identities
- Session save/load
- Leak checks
- Live console
Pro
8 persistent identity slots
For teams running multiple agents across accounts, workflows, and approvals.
$99/mo
- 8 saved browser identities
- Session cloning
- Opt-in personality controls
- Web3 wallet monitoring
Fleet
20 dedicated persistent identity slots
For production agent fleets that need dedicated slots and a shared supervision surface.
$299/mo
- 20 dedicated slots
- Fleet console
- Structured event stream
- Priority onboarding
Plans are sold by persistent identity slot, not per-browser-hour. Every plan starts with a 14-day card-on-file trial. No free tier.
MCP-native by default