# vercel-labs/agent-browser architecture

> Architecture diagram of the GitHub repository vercel-labs/agent-browser, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-16). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/vercel-labs/agent-browser
- Repository: https://github.com/vercel-labs/agent-browser

## Overview

agent-browser is a Rust browser-automation CLI for AI agents, with typed MCP access and live browser streaming. The main workflow is an agent invoking CLI commands to launch/connect to a browser, perform page actions, and receive snapshots or other results. The README establishes the user-facing capabilities; source-index references and sampled excerpts support selected internal links. Unsampled subsystems are represented where documented, but their internal call paths are not inferred.

Main entry points: [`cli/src/main.rs`](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/main.rs) (CLI entry), [`cli/src/commands.rs`](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/commands.rs) (Command parsing), [`cli/src/native/cdp/client.rs`](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/cdp/client.rs) (CDP transport), [`cli/src/native/browser.rs`](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/browser.rs) (Browser control)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_entry["CLI entry"]
  node_cli["CLI entry<br/>[main.rs]"]
  node_command_parse["Command parsing<br/>[commands.rs]"]
  node_flags["Flags and config<br/>[flags.rs]"]
end

subgraph group_automation["Browser automation"]
  node_browser["Browser control<br/>[browser.rs]"]
  node_actions["Page actions<br/>[actions.rs]"]
  node_cdp["CDP transport<br/>[client.rs]"]
  node_snapshot["Accessibility snapshot<br/>[snapshot.rs]"]
  node_network["Network control<br/>[network.rs]"]
  node_webmcp["WebMCP tools<br/>[webmcp.rs]"]
  node_read["Readable page content<br/>[read.rs]"]
  node_recording["Session recording<br/>[recording.rs]"]
end

subgraph group_interfaces["Agent interfaces"]
  node_mcp["MCP server<br/>[mcp.rs]"]
  node_skills["Agent skills<br/>[skills.rs]"]
  node_stream["Live streaming<br/>[mod.rs]"]
  node_stream_loop["CDP stream loop<br/>[cdp_loop.rs]"]
  node_chat["Browser chat<br/>[chat.rs]"]
  node_eve_extension["Eve extension tools<br/>[drag.ts]"]
end

subgraph group_operations["Runtime operations"]
  node_install["Browser installation<br/>[install.rs]"]
  node_doctor["Environment diagnostics<br/>[mod.rs]"]
end

subgraph group_dashboard["Session dashboard"]
  node_dashboard["Dashboard UI<br/>[page.tsx]"]
  node_dashboard_routes["Dashboard routes"]
end

node_agent(("AI agent"))
node_browser_service(("Browser runtime"))

node_agent -->|"runs commands"| node_cli
node_cli -->|"dispatches commands"| node_command_parse
node_cli -->|"parses options"| node_flags
node_cli -->|"serves MCP"| node_mcp
node_cli -->|"installs browser"| node_install
node_cli -->|"runs checks"| node_doctor
node_command_parse -->|"parses tools"| node_webmcp
node_command_parse -->|"parses read"| node_read
node_command_parse -->|"parses recording"| node_recording
node_agent -->|"controls browser"| node_browser
node_browser -->|"drives browser"| node_browser_service
node_agent -->|"requests snapshot"| node_snapshot
node_agent -->|"requests actions"| node_actions
node_agent -->|"configures network"| node_network
node_agent -->|"discovers tools"| node_mcp
node_agent -->|"loads guidance"| node_skills
node_agent -->|"enables streaming"| node_stream
node_stream -->|"uses loop"| node_stream_loop
node_agent -->|"starts chat"| node_chat
node_cdp -.->|"sends protocol"| node_browser_service
node_agent -->|"installs Chrome"| node_install
node_agent -.->|"checks setup"| node_doctor
node_agent -.->|"views sessions"| node_dashboard
node_dashboard -->|"uses routes"| node_dashboard_routes
node_agent -.->|"invokes extension"| node_eve_extension
node_cdp -.->|"returns protocol data"| node_cli

click node_cli "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/main.rs"
click node_command_parse "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/commands.rs"
click node_flags "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/flags.rs"
click node_browser "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/browser.rs"
click node_actions "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/actions.rs"
click node_cdp "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/cdp/client.rs"
click node_snapshot "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/snapshot.rs"
click node_network "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/network.rs"
click node_webmcp "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/webmcp.rs"
click node_read "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/read.rs"
click node_mcp "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/mcp.rs"
click node_skills "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/skills.rs"
click node_stream "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/stream/mod.rs"
click node_stream_loop "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/stream/cdp_loop.rs"
click node_chat "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/chat.rs"
click node_recording "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/recording.rs"
click node_install "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/install.rs"
click node_doctor "https://github.com/vercel-labs/agent-browser/blob/main/cli/src/doctor/mod.rs"
click node_dashboard "https://github.com/vercel-labs/agent-browser/blob/main/packages/dashboard/src/app/page.tsx"
click node_dashboard_routes "https://github.com/vercel-labs/agent-browser/blob/main/packages/dashboard/src/lib/dashboard-routes.ts"
click node_eve_extension "https://github.com/vercel-labs/agent-browser/blob/main/packages/%40agent-browser/eve/extension/tools/drag.ts"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_cli,node_command_parse,node_flags,node_browser_service toneBlue
class node_browser,node_actions,node_cdp,node_snapshot,node_network,node_webmcp,node_read,node_recording toneAmber
class node_mcp,node_skills,node_stream,node_stream_loop,node_chat,node_eve_extension toneMint
class node_install,node_doctor toneRose
class node_dashboard,node_dashboard_routes,node_agent toneIndigo
```

## Components

### CLI entry

- [CLI entry](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/main.rs): `cli/src/main.rs`
- [Command parsing](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/commands.rs): `cli/src/commands.rs`
- [Flags and config](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/flags.rs): `cli/src/flags.rs`

### Browser automation

- [Browser control](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/browser.rs): `cli/src/native/browser.rs`
- [Page actions](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/actions.rs): `cli/src/native/actions.rs`
- [CDP transport](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/cdp/client.rs): `cli/src/native/cdp/client.rs`
- [Accessibility snapshot](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/snapshot.rs): `cli/src/native/snapshot.rs`
- [Network control](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/network.rs): `cli/src/native/network.rs`
- [WebMCP tools](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/webmcp.rs): `cli/src/native/webmcp.rs`
- [Readable page content](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/read.rs): `cli/src/read.rs`
- [Session recording](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/recording.rs): `cli/src/native/recording.rs`

### Agent interfaces

- [MCP server](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/mcp.rs): `cli/src/mcp.rs`
- [Agent skills](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/skills.rs): `cli/src/skills.rs`
- [Live streaming](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/stream/mod.rs): `cli/src/native/stream/mod.rs`
- [CDP stream loop](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/native/stream/cdp_loop.rs): `cli/src/native/stream/cdp_loop.rs`
- [Browser chat](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/chat.rs): `cli/src/chat.rs`
- [Eve extension tools](https://github.com/vercel-labs/agent-browser/blob/main/packages/%40agent-browser/eve/extension/tools/drag.ts): `packages/@agent-browser/eve/extension/tools/drag.ts`

### Runtime operations

- [Browser installation](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/install.rs): `cli/src/install.rs`
- [Environment diagnostics](https://github.com/vercel-labs/agent-browser/blob/main/cli/src/doctor/mod.rs): `cli/src/doctor/mod.rs`

### Session dashboard

- [Dashboard UI](https://github.com/vercel-labs/agent-browser/blob/main/packages/dashboard/src/app/page.tsx): `packages/dashboard/src/app/page.tsx`
- [Dashboard routes](https://github.com/vercel-labs/agent-browser/blob/main/packages/dashboard/src/lib/dashboard-routes.ts): `packages/dashboard/src/lib/dashboard-routes.ts`

### Other components

- **AI agent**
- **Browser runtime**

## Connections

- AI agent → CLI entry: runs commands (evidence: `README.md`)
- CLI entry → Command parsing: dispatches commands (evidence: `cli/src/main.rs`)
- CLI entry → Flags and config: parses options (evidence: `cli/src/main.rs`)
- CLI entry → MCP server: serves MCP (evidence: `cli/src/main.rs`)
- CLI entry → Browser installation: installs browser (evidence: `cli/src/main.rs`)
- CLI entry → Environment diagnostics: runs checks (evidence: `cli/src/main.rs`)
- Command parsing → WebMCP tools: parses tools (evidence: `cli/src/commands.rs`)
- Command parsing → Readable page content: parses read (evidence: `cli/src/commands.rs`)
- Command parsing → Session recording: parses recording (evidence: `cli/src/commands.rs`)
- AI agent → Browser control: controls browser (evidence: `README.md`)
- Browser control → Browser runtime: drives browser (evidence: `README.md`)
- AI agent → Accessibility snapshot: requests snapshot (evidence: `README.md`)
- AI agent → Page actions: requests actions (evidence: `README.md`)
- AI agent → Network control: configures network (evidence: `README.md`)
- AI agent → MCP server: discovers tools (evidence: `README.md`)
- AI agent → Agent skills: loads guidance
- AI agent → Live streaming: enables streaming (evidence: `README.md`)
- Live streaming → CDP stream loop: uses loop (evidence: `cli/src/native/stream/mod.rs`)
- AI agent → Browser chat: starts chat (evidence: `README.md`)
- CDP transport → Browser runtime: sends protocol
- AI agent → Browser installation: installs Chrome (evidence: `README.md`)
- AI agent → Environment diagnostics: checks setup
- AI agent → Dashboard UI: views sessions
- Dashboard UI → Dashboard routes: uses routes (evidence: `packages/dashboard/src/app/page.tsx`)
- AI agent → Eve extension tools: invokes extension
- CDP transport → CLI entry: returns protocol data
