Quick Start

Prerequisites

Before starting with OpenViking, ensure your environment meets the following requirements:

Python 3.10+

Python 3.10 or higher is required

Network Access

Stable connection for dependencies and model services

AI Model Access

VLM and embedding model API credentials

Operating System

Linux, macOS, or Windows

Advanced Requirements (only for building from source):

Go 1.22+ for AGFS components
GCC 9+ or Clang 11+ for C++ extensions

Installation

Install OpenViking

Install the Python package using pip:

pip install openviking --upgrade --force-reinstall

Using uv for faster installation:

uv pip install openviking --upgrade

Verify Installation

Check that OpenViking is installed correctly:

python -c "import openviking; print(openviking.__version__)"

Optional: Rust CLI

For advanced users, OpenViking provides a high-performance Rust CLI:

# Quick install
curl -fsSL https://raw.githubusercontent.com/volcengine/OpenViking/main/crates/ov_cli/install.sh | bash

# Or build from source
cargo install --git https://github.com/volcengine/OpenViking ov_cli

Model Configuration

OpenViking requires two types of models:

VLM Model

For image and content understanding

Embedding Model

For vectorization and semantic retrieval

Supported VLM Providers

OpenViking supports three VLM providers:

Volcengine (Doubao)
OpenAI
LiteLLM

Recommended - Cost-effective with good performance, free quota for new users.

{
  "vlm": {
    "provider": "volcengine",
    "model": "doubao-seed-2-0-pro-260215",
    "api_key": "your-api-key",
    "api_base": "https://ark.cn-beijing.volces.com/api/v3"
  }
}

You can also use endpoint IDs:

{
  "vlm": {
    "provider": "volcengine",
    "model": "ep-20241220174930-xxxxx",
    "api_key": "your-api-key",
    "api_base": "https://ark.cn-beijing.volces.com/api/v3"
  }
}

Get your API key from Volcengine ARK Console

Use OpenAI’s official API or OpenAI-compatible endpoints:

{
  "vlm": {
    "provider": "openai",
    "model": "gpt-4o",
    "api_key": "your-api-key",
    "api_base": "https://api.openai.com/v1"
  }
}

Get your API key from OpenAI Platform

Unified access to various models including Anthropic, DeepSeek, Gemini, Qwen, vLLM, Ollama, and more.Anthropic (Claude):

{
  "vlm": {
    "provider": "litellm",
    "model": "claude-3-5-sonnet-20240620",
    "api_key": "your-anthropic-api-key"
  }
}

Qwen (DashScope):

{
  "vlm": {
    "provider": "litellm",
    "model": "dashscope/qwen-turbo",
    "api_key": "your-dashscope-api-key",
    "api_base": "https://dashscope.aliyuncs.com/compatible-mode/v1"
  }
}

Local Models (Ollama):

{
  "vlm": {
    "provider": "litellm",
    "model": "ollama/llama3.1",
    "api_base": "http://localhost:11434"
  }
}

The system auto-detects common models like claude-*, deepseek-*, gemini-*, ollama/*, etc. See LiteLLM Providers for complete list.

Configuration File

Create Configuration Directory

mkdir -p ~/.openviking

Create Configuration File

Create ~/.openviking/ov.conf with your model settings:

{
  "storage": {
    "workspace": "/home/your-name/openviking_workspace"
  },
  "log": {
    "level": "INFO",
    "output": "stdout"
  },
  "embedding": {
    "dense": {
      "api_base": "https://ark.cn-beijing.volces.com/api/v3",
      "api_key": "your-embedding-api-key",
      "provider": "volcengine",
      "dimension": 1024,
      "model": "doubao-embedding-vision-250615"
    },
    "max_concurrent": 10
  },
  "vlm": {
    "api_base": "https://ark.cn-beijing.volces.com/api/v3",
    "api_key": "your-vlm-api-key",
    "provider": "volcengine",
    "model": "doubao-seed-2-0-pro-260215",
    "max_concurrent": 100
  }
}

Replace your-name, your-embedding-api-key, and your-vlm-api-key with your actual values.

Set Environment Variable (Optional)

If your config file is not at the default location:

export OPENVIKING_CONFIG_FILE=~/.openviking/ov.conf

Your First Example

Let’s create a complete example that demonstrates OpenViking’s core features.

Create Python Script

Create a file named example.py:

example.py

import openviking as ov

# Initialize OpenViking client with data directory
client = ov.OpenViking(path="./data")

try:
    # Initialize the client
    client.initialize()
    print("✓ Client initialized")
    
    # Add resource (supports URL, file, or directory)
    add_result = client.add_resource(
        path="https://raw.githubusercontent.com/volcengine/OpenViking/refs/heads/main/README.md"
    )
    root_uri = add_result['root_uri']
    print(f"✓ Resource added: {root_uri}")
    
    # Explore the resource tree structure
    ls_result = client.ls(root_uri)
    print(f"\n📁 Directory structure:\n{ls_result}")
    
    # Use glob to find markdown files
    glob_result = client.glob(pattern="**/*.md", uri=root_uri)
    if glob_result['matches']:
        first_file = glob_result['matches'][0]
        content = client.read(first_file)
        print(f"\n📄 Content preview ({first_file}):\n{content[:200]}...")
    
    # Wait for semantic processing to complete
    print("\n⏳ Waiting for semantic processing...")
    client.wait_processed()
    print("✓ Processing complete")
    
    # Get abstract and overview of the resource (L0 and L1 layers)
    abstract = client.abstract(root_uri)
    overview = client.overview(root_uri)
    print(f"\n📝 L0 Abstract:\n{abstract}")
    print(f"\n📋 L1 Overview:\n{overview[:500]}...")
    
    # Perform semantic search
    print("\n🔍 Semantic search: 'what is openviking'")
    results = client.find("what is openviking", target_uri=root_uri)
    print("\nSearch results:")
    for r in results.resources:
        print(f"  • {r.uri} (score: {r.score:.4f})")
    
    # Close the client
    client.close()
    print("\n✓ Client closed successfully")
    
except Exception as e:
    print(f"❌ Error: {e}")
    import traceback
    traceback.print_exc()

Run the Script

python example.py

Expected Output

You should see output similar to:

✓ Client initialized
✓ Resource added: viking://resources/github.com/volcengine/OpenViking/README.md

📁 Directory structure:
[...]

📄 Content preview:
<div align="center">

<picture>
  <img alt="OpenViking" src="docs/images/banner.jpg" width="100%" height="auto">
</picture>...

⏳ Waiting for semantic processing...
✓ Processing complete

📝 L0 Abstract:
OpenViking is an open-source context database for AI Agents...

📋 L1 Overview:
# Overview
OpenViking unifies the management of context through a file system paradigm...

🔍 Semantic search: 'what is openviking'

Search results:
  • viking://resources/github.com/volcengine/OpenViking/README.md (score: 0.8523)

✓ Client closed successfully

Congratulations! You’ve successfully:

Added a resource to OpenViking
Explored the filesystem structure
Retrieved hierarchical context (L0/L1 layers)
Performed semantic search

Key API Operations

Here are the essential operations you’ll use with OpenViking:

Resource Management

# Add resources (URL, local file, or directory)
client.add_resource(path="https://example.com/doc.pdf")
client.add_resource(path="/path/to/local/file.md")
client.add_resource(path="/path/to/directory/")

# Wait for semantic processing
client.wait_processed()

# Remove resources
client.rm("viking://resources/example/", recursive=True)

Filesystem Operations

# List directory contents
client.ls("viking://resources/")

# Read file content
content = client.read("viking://resources/doc.md")

# Create directory
client.mkdir("viking://agent/skills/custom/")

# Find files by pattern
matches = client.glob("**/*.py", uri="viking://resources/project/")

# Tree view
tree = client.tree("viking://resources/", max_depth=2)

Hierarchical Context Access

# Get L0 abstract (~100 tokens)
abstract = client.abstract("viking://resources/project/")

# Get L1 overview (~2k tokens)
overview = client.overview("viking://resources/project/")

# Get L2 full content
content = client.read("viking://resources/project/docs/api.md")

Semantic Search

# Search across all resources
results = client.find("user authentication", limit=10)

# Search within specific URI
results = client.find(
    query="API endpoints",
    target_uri="viking://resources/project/",
    limit=5
)

# Access results
for resource in results.resources:
    print(f"{resource.uri} - Score: {resource.score}")
    print(f"Content: {resource.content[:200]}...")

Session Management

# Create a session
session = client.create_session()
session_id = session["session_id"]

# Add messages
client.add_message(session_id, "user", "How do I configure OpenViking?")
client.add_message(session_id, "assistant", "To configure OpenViking...")

# Get session info
info = client.get_session(session_id)

# List all sessions
sessions = client.list_sessions()

# Commit session (extract memories)
client.commit_session(session_id)

Next Steps

Server Deployment

Learn how to deploy OpenViking as a production HTTP service

API Reference

Explore the complete API documentation

Configuration Guide

Advanced configuration options and model providers

Examples

Browse code examples and integration patterns

Troubleshooting

Import Error: Cannot find module 'openviking'

Make sure OpenViking is installed in your current Python environment:

pip install openviking --upgrade
python -c "import openviking; print(openviking.__version__)"

Configuration file not found

Ensure your config file exists at ~/.openviking/ov.conf or set the OPENVIKING_CONFIG_FILE environment variable:

export OPENVIKING_CONFIG_FILE=/path/to/your/ov.conf

API authentication errors

Verify your API keys are correct in the configuration file. For Volcengine, check ARK Console. For OpenAI, check OpenAI Platform.

Semantic processing takes too long

Semantic processing time depends on:

Resource size and complexity
Model API response time
Network latency

You can check processing status:

status = client.wait_processed()
print(status)

Get Started

Core Concepts

Guides

Integrations

Prerequisites

Python 3.10+

Network Access

AI Model Access

Operating System

Installation

Optional: Rust CLI

Model Configuration

VLM Model

Embedding Model

Supported VLM Providers

Configuration File

Your First Example

Key API Operations

Resource Management

Filesystem Operations

Hierarchical Context Access

Semantic Search

Session Management

Next Steps

Server Deployment

API Reference

Configuration Guide

Examples

Troubleshooting

Get Started

Core Concepts

Guides

Integrations

​Prerequisites

Python 3.10+

Network Access

AI Model Access

Operating System

​Installation

​Optional: Rust CLI

​Model Configuration

VLM Model

Embedding Model

​Supported VLM Providers

​Configuration File

​Your First Example

​Key API Operations

​Resource Management

​Filesystem Operations

​Hierarchical Context Access

​Semantic Search

​Session Management

​Next Steps

Server Deployment

API Reference

Configuration Guide

Examples

​Troubleshooting

Prerequisites

Installation

Optional: Rust CLI

Model Configuration

Supported VLM Providers

Configuration File

Your First Example

Key API Operations

Resource Management

Filesystem Operations

Hierarchical Context Access

Semantic Search

Session Management

Next Steps

Troubleshooting