superbizon007/linux-chrome-agent repository overview

git clone <repo>
cd apps/linux_chrome_agent
docker build --build-context a2a_pkg=../../packages/a2a_agent -f Dockerfile.ubuntu -t linux_chrome_agent:ubuntu .

docker run -d \
  --name chrome_agent \
  -p 443:443 \
  -p 6901:6901 \
  -p 5901:5901 \
  -p 9222:9222 \
  -e VLM_URL=https://api.openai.com/v1 \
  -e VLM_API_KEY=sk-... \
  -e VLM_MODEL=gpt-4o \
  linux_chrome_agent:ubuntu --wait

docker build --build-context a2a_pkg=../../packages/a2a_agent -f Dockerfile.dev -t linux_chrome_agent_dev:ubuntu .

docker run -d \
  --name chrome_agent_dev \
  -p 6901:6901 \
  -p 5901:5901 \
  -p 9222:9222 \
  linux_chrome_agent_dev:ubuntu --wait

VLM_URL=https://api.openai.com/v1 \
VLM_API_KEY=sk-... \
VLM_MODEL=gpt-4o \
CDP_HOST=localhost \
CDP_PORT=9222 \
uv run python -m app.server

API_KEY=your-token \
VLM_URL=https://api.openai.com/v1 \
VLM_API_KEY=sk-... \
VLM_MODEL=gpt-4o \
CDP_HOST=localhost \
CDP_PORT=9222 \
uv run python -m app.server

curl -sk -X POST https://localhost/ \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": "1",
    "method": "message/send",
    "params": {
      "id": "task-1",
      "message": {
        "role": "user",
        "parts": [{"text": "Go to example.com and tell me the page title"}]
      }
    }
  }'

curl -sk -X POST https://localhost/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "browser-agent",
    "messages": [{"role": "user", "content": "Go to example.com and tell me the page title"}]
  }'

curl -sk https://localhost/.well-known/agent.json

$TRAJECTORY_DIR/
  20260225_181523/              # run timestamp (YYYYMMDD_HHMMSS)
    summary.md                  # task, result, reusable plan, and cost summary
    turn_0000/
      screenshot.png            # browser state at start of step
      screenshot_marked.png     # annotated with detected elements (DOM mode)
      screenshot_annotated.png  # annotated with the proposed click target
      elements.json             # detected interactive elements
      llm_0_request.json        # messages sent to the VLM
      llm_0_response.json       # raw VLM reply + parsed action/args/reasoning
      locator.json              # locator model request + response (when LOCATOR_URL is set)
    turn_0001/
      ...
  20260225_182041/              # next run
    ...

docker run -d \
  -e VLM_URL=... -e VLM_API_KEY=... \
  -e TRAJECTORY_DIR=/data/trajectory \
  -v /tmp/agent-logs:/data/trajectory \
  linux_chrome_agent:ubuntu --wait

# Default — start Chrome, VNC, and API server, keep running
docker run linux_chrome_agent:ubuntu --wait

# Headless mode — no VNC, CDP + API only
docker run linux_chrome_agent:ubuntu --no-vnc

# Debug — verbose startup output
docker run linux_chrome_agent:ubuntu --debug

# Skip startup — run a custom command instead
docker run linux_chrome_agent:ubuntu --skip bash

┌───────────────────────────────────────────────────┐
│                 Docker Container                   │
│                                                   │
│  Chrome (CDP :9223) ◄─ socat ◄─ :9222 (ext)      │
│         │                                         │
│  CDPClient          Network idle tracking         │
│         │           New-tab detection             │
│  Agent (browser.py)                               │
│         │                                         │
│  LangGraph (graph.py)                             │
│    take_screenshot                                │
│       │   OmniParser (optional)                   │
│       │   Popup detection                         │
│    call_vlm ──► VLM (OpenAI-compatible)           │
│    execute_action                                 │
│       │   Locator model (optional)                │
│       │   Network idle wait                       │
│       │   Loop detection                          │
│    check_updates ── drain TaskInbox               │
│       │             handle pause/resume           │
│       ├── (loop) → take_screenshot                │
│       ├── (done) → generate_summary ──► END       │
│       └── (max steps) ──► END                     │
│                                                   │
│  FastAPI Server (:443 HTTPS)                      │
│    ├── A2A router  (POST /)                       │
│    └── OpenAI router (/v1/...)                    │
└───────────────────────────────────────────────────┘