🐳 Docker Deployment Now Available

VibeSurf can now run in Docker containers for cloud deployment and enhanced security. The containerized environment acts as a sandbox, making it safer and more convenient to use.

Quick Start

Option 1: Using docker-compose (Recommended)

# 1.  Clone VibeSurf Repo
git clone https://github.com/vibesurf-ai/VibeSurf
#  Optional: Edit docker-compose.yml to modify envs

# 2. Start VibeSurf
docker-compose up -d

# 3. Access VibeSurf
# - Backend: http://localhost:9335
# - Browser VNC (Web): http://localhost:6080 (default password: vibesurf)

Option 2: Using docker run

Pull the image

docker pull ghcr.io/vibesurf-ai/vibesurf:latest

Run the container

docker run --name vibesurf -d --restart unless-stopped \
-p 9335:9335 \
-p 6080:6080 \
-p 5901:5901 \
-v ./data:/data \
-e IN_DOCKER=true \
-e VIBESURF_WORKSPACE=/data/vibesurf_workspace \
-e RESOLUTION=1440x900x24 \
-e VNC_PASSWORD=vibesurf \
--shm-size=4g \
--cap-add=SYS_ADMIN \
ghcr.io/vibesurf-ai/vibesurf:latest

View browser in docker via VNC: http://127.0.0.1:6080/

Or you can use claude code or codex to control vibesurf browser and real-time preview in vscode, following below:

Open the Internal Browser

• Open the Command Palette:

  Windows/Linux: Ctrl + Shift + P
  macOS: Cmd + Shift + P
  

• Type and select: Simple Browser: Show
• Enter the URL: http://127.0.0.1:6080/
• Press Enter.

Adjust VNC Screen Size

• Click the Settings icon (gear button) on the left sidebar of the VNC viewer.
• Under Scaling Mode, select Local Scaling.
• This will allow the VNC screen to automatically resize to fit the window.

What's Changed

• Fix BrowserEvaluateJavaScript component multiline input escape sequen… by @vvincent1234 in #145
• Update README video section with YouTube tutorial link by @vvincent1234 in #146

Full Changelog: 0.5.7...0.5.8

• Add start_console_logging and stop_console_logging actions to monitor browser console output
• Add start_network_logging and stop_network_logging actions to capture HTTP traffic
• Console logs saved to both .log (human-readable) and .json (structured) formats
• Network logs saved as HAR (HTTP Archive) format with full request/response details
• Add Fetch content from URL
• Fix video subtitle

Full Changelog: 0.4.8...0.4.9

VibeSurf integration for Claude Code - Control real browsers, execute AI skills, run workflows, and integrate with external apps.

https://github.com/vibesurf-ai/claude-surf

# Add the marketplace
/plugin marketplace add vibesurf-ai/claude-surf

# Install the plugin
/plugin install surf

What You Can Do

• Browser Control - Navigate websites, interact with elements
• AI Skills - Search, crawl, extract data, summarize content
• Workflows - Execute pre-built automation workflows
• App Integrations - Gmail, GitHub, Slack, and 100+ apps via Composio/MCP
• Browser-Use Agent - Parallel multi-task automation using AI sub-agents

Happy New Year! 🎉🎉

In this version, we expose all VibeSurf actions as HTTP REST APIs, enabling programmatic control of browser automation, AI skills, and workflow execution.

Core Endpoints

Search Actions

GET /api/tool/search?keyword=browser

Returns available actions with descriptions. Filter by keyword or list all.

Get Action Parameters

GET /api/tool/{action_name}/params

Returns JSON schema for action parameters.

Execute Action

POST /api/tool/execute
Content-Type: application/json

{
  "action_name": "execute_browser_use_agent",
  "action_params": {
    "tasks": [
      {"task": "Open VibeSurf official website"},
      {"task": "Open VibeSurf GitHub repo"}
    ]
  }
}

What You Can Do

Browser Control

• Atomic operations: browser.click_element, browser.input_text, browser.scroll_down
• Browser state: browser.get_current_state, browser.get_all_tabs
• Browser task execution via execute_browser_use_agent

AI Skills

• advanced_search - Intelligent web search with LLM
• crawl - Extract structured data from websites
• code - Execute JavaScript in browser context
• summary - Summarize content with AI
• type / to find more

Integration Tools

• Composio toolkit support via extra_tool parameter
• Control Gmail, Google Calendar, GitHub, and 100+ apps

Workflow Execution

• Execute pre-built workflow templates via action API
• Combine deterministic automation with AI intelligence

Example Usage

Execute parallel browser tasks:

curl -X POST 'http://127.0.0.1:9335/api/tool/execute' \
  -H 'Content-Type: application/json' \
  -d '{
    "action_name": "execute_browser_use_agent",
    "action_params": {
      "tasks": [
        {"task": "Search for VibeSurf documentation"},
        {"task": "Open GitHub trending page"}
      ]
    }
  }'

Perform advanced web search:

curl -X POST 'http://127.0.0.1:9335/api/tool/execute' \
  -H 'Content-Type: application/json' \
  -d '{
    "action_name": "advanced_search",
    "action_params": {
      "query": "latest AI browser automation tools"
    }
  }'

Full Changelog: 0.3.15...0.3.16

自取：https://vibe-surf.com/workflows/352f08a6-df04-422a-b1e5-d139f5ccb2b6

Full Changelog: 0.3.9...0.3.10

What's Changed

• Phone by @vvincent1234 in #120
• Simplify installation and add development guide by @vvincent1234 in #121
• Add GLM (ZhipuAI) model support by @vvincent1234 in #122

Full Changelog: 0.3.5...0.3.6

Convert any browser workflow into reusable skills for browser agents, combining the deterministic precision and speed of workflows with the intelligence of AI agents.

Even OpenAI is adopting Claude skills - browser workflow skills are the inevitable next step. VibeSurf is positioning ahead of the curve by solving the core efficiency problem plaguing AI browsers.

🎯 Why This Changes Everything

The Problem with Current AI Browsers

• Low success rates in task execution
• Inefficient performance with high token consumption
• Repetitive agent calls for routine, predictable tasks
• Leading AI browsers (Comet, Dia, ChatGPT Atlas) struggle with these limitations

Our Solution: Workflow-Agent Hybrid

• Fixed processes → Use fast, accurate workflows
• Dynamic decisions → Deploy agent intelligence only when needed
• Best of both worlds → Deterministic reliability + AI flexibility

💡 How It Works

Setup Process:

• Access Settings → Click top-right corner settings
• Navigate to Workflows → Select "Workflow" tab
• Enable Skills → Click "Add to Skill" button on workflow cards
• Configure Inputs (Optional) → Set "expose workflow inputs" to allow dynamic parameter modification
• Confirm → Skills are now available to your browser agent

Usage:

• Automatic Discovery → Agent automatically finds and uses available workflow skills
• Manual Selection → Type @flow to browse and select specific workflow skills
• Dynamic Execution → Agent adapts workflow parameters based on context

A workflow that helps you download any social media video , ASR , transcribe and add subtitle to video

1. 优化windows 安装包，更好支持中国地区的安装
2. Fix start browser timeout.
3. 修复小红书API
4. 增加知乎的website api
5. 优化loop component
6. Fix browser-use bugs
7. Add filesystem compoenents
8. Speed up getting weather and newsnow

Full Changelog: 0.2.46...0.2.47