Installation

CLI Install (Fastest)

The quickest way to set up DeepSyte for a new agent is with the CLI. Use the one-command onboarding flow:

npx deepsyte setup --client cursor    # or: vscode, windsurf, claude, claude-code

This will:

1. Open your browser to authenticate via OAuth if needed
2. Configure the MCP server in your chosen client
3. Install or repair the managed core DeepSyte skill in ~/.agents/skills/deepsyte, including workflows/sitewide-performance-audit/WORKFLOW.md

Use this path when you want the fastest way to get browser truth into a fresh IDE, CLI, or MCP workflow.

If you prefer to do onboarding in two steps, you can still run:

npx deepsyte login
npx deepsyte install cursor    # or: vscode, windsurf, claude, claude-code

That two-step path reaches the same result as setup --client <client> for most clients. The main nuances are that install vscode writes a workspace-local .vscode/mcp.json, while install claude-code prints the claude mcp add ... command for you to run manually.

The first packaged workflow is ~/.agents/skills/deepsyte/workflows/sitewide-performance-audit/WORKFLOW.md for repeatable multi-page performance audits.

deepsyte skills ... only manages that local core DeepSyte skill. To discover or install community skills from the wider agent ecosystem, use the find-skills workflow or npx skills find ... / npx skills add ....

You can also install it globally for ongoing use:

npm install -g deepsyte
deepsyte --help

For public sites, the CLI now mirrors the MCP remote browser workflow directly. Start a remote session, keep the returned sessionId, and continue with browse:* commands:

deepsyte browse https://example.com
deepsyte browse:click <sessionId> "Sign in"
deepsyte browse:console <sessionId> --level error
deepsyte browse:network-errors <sessionId>
deepsyte browse:a11y <sessionId>
deepsyte browse:perf <sessionId>
deepsyte browse:seo <sessionId>
deepsyte browse:close <sessionId>
deepsyte auth:test https://example.com
deepsyte auth:find-login https://example.com
deepsyte auth:smart-login https://example.com/sign-in --username user@example.com --password secret

For extension-free local browser workflows, you can also launch an installed browser in a dedicated DeepSyte profile:

deepsyte browser open https://example.com
deepsyte browser open https://example.com --record-video
deepsyte browser status
deepsyte browser goto https://example.org
deepsyte browser back
deepsyte browser forward
deepsyte browser click-at 320 480
deepsyte browser hover ".menu-trigger"
deepsyte browser wait-for ".results-loaded" --timeout 8000
deepsyte browser select "select[name=country]" "Australia"
deepsyte browser viewport 393 852
deepsyte browser screenshot
deepsyte browser console --level error
deepsyte browser network-errors
deepsyte browser network-requests --resource-type fetch --min-duration 200
deepsyte browser evidence --label checkout-bug
deepsyte browser close --evidence --label checkout-bug
deepsyte browser cookies get
deepsyte browser storage getAll --type localStorage
deepsyte browser eval "document.title"
deepsyte browser a11y --max-depth 6
deepsyte browser perf
deepsyte browser seo
deepsyte browser close

The CLI explains why local access is needed, asks for explicit approval before it opens the browser, continuously captures console logs and network activity while that managed browser stays open, and then reconnects to that tracked session for follow-up local actions including history navigation, coordinate clicks, hover states, wait conditions, dropdown selection, viewport changes, cookies/storage inspection, script evaluation, accessibility inspection, performance metrics, SEO auditing, and timestamped evidence bundle export. Use browser evidence for a live proof snapshot, or browser close --evidence when you want the finalized local .webm included in the same bundle after recording finishes.

Use remote browser workflows for public sites. Escalate to the managed local browser when the work requires localhost, intranet access, VPN-only apps, stronger realism, or explicit human approval.

VS Code Extension Preview

A native DeepSyte VS Code extension is now being developed in the monorepo for a dedicated Activity Bar sidebar, automatic browser OAuth sign-in, automatic editor MCP setup, automatic managed core skill sync, API key fallback, command palette actions, output logs, a live activity timeline panel, and native MCP registration.

Current preview commands include:

• DeepSyte: Sign In
• DeepSyte: Check Status
• DeepSyte: Take Screenshot
• DeepSyte: Open Timeline
• DeepSyte: Configure Editor Integration
• DeepSyte: Sync Core Skill

The sidebar also surfaces quick actions and recent activity directly inside VS Code, and the extension opens the browser sign-in flow automatically when no credentials are stored. After sign-in, the extension also configures the matching MCP integration automatically when needed and repairs the managed core skill in ~/.agents/skills/deepsyte.

Until the Marketplace release is ready, install the preview VSIX and sign in once to let the extension configure the editor automatically. If you need to repair the setup later, use DeepSyte: Configure Editor Integration for the MCP connection or DeepSyte: Sync Core Skill for the managed local skill.

Manual Configuration

DeepSyte works with any client that supports the Model Context Protocol (MCP). Here's how to configure each one manually.

If you are setting up a new agent and want the managed core skill plus packaged workflows installed automatically, prefer npx deepsyte setup --client <client> before falling back to the manual snippets below.

Cursor

Open Cursor Settings → MCP Servers, or edit ~/.cursor/mcp.json directly:

{
  "mcpServers": {
    "deepsyte": {
      "url": "https://deepsyte-api-production.up.railway.app/mcp/YOUR_API_KEY"
    }
  }
}

Windsurf

Open Windsurf Settings → MCP, or edit the config file:

{
  "mcpServers": {
    "deepsyte": {
      "serverUrl": "https://deepsyte-api-production.up.railway.app/mcp/YOUR_API_KEY"
    }
  }
}

Claude Desktop

Edit claude_desktop_config.json (found in ~/Library/Application Support/Claude/ on macOS or %APPDATA%\Claude\ on Windows):

{
  "mcpServers": {
    "deepsyte": {
      "url": "https://deepsyte-api-production.up.railway.app/mcp/YOUR_API_KEY"
    }
  }
}

VS Code (Copilot)

Add to your VS Code settings (.vscode/mcp.json):

{
  "mcpServers": {
    "deepsyte": {
      "url": "https://deepsyte-api-production.up.railway.app/mcp/YOUR_API_KEY"
    }
  }
}

This file is workspace-local, so repeat this step for each workspace that should have DeepSyte enabled.

Any MCP Client

The server endpoint is:

https://deepsyte-api-production.up.railway.app/mcp/YOUR_API_KEY

It supports Streamable HTTP transport — compatible with all MCP clients that support HTTP-based servers.

Verify Installation

After configuring, ask your AI assistant:

Open https://example.com, inspect the page, and tell me what you see

Then try a verification-oriented task:

Test my sign-in flow and capture proof if anything fails

If you get browser output and evidence-rich results back, you're all set.

Available Tools

Once connected, your AI assistant has access to 46+ tools: