Deploy DeepWiki with Docker

Docker is the fastest way to run DeepWiki Open. A single image bundles the Python API server and the Next.js frontend together, so you only need to supply your API keys and run one command. All cloned repositories, embeddings, and cached wiki content are stored in ~/.adalflow on your host machine, so your data persists across container restarts.

Docker Compose
Docker Run
Build locally

Docker Compose is the recommended approach. It reads your API keys from a .env file and handles port mapping, volume mounts, and health checks automatically.

Clone the repository

git clone https://github.com/AsyncFuncAI/deepwiki-open.git
cd deepwiki-open

Create your .env file

Create a .env file in the project root with your API keys:

.env

GOOGLE_API_KEY=your_google_api_key
OPENAI_API_KEY=your_openai_api_key
# Optional: Use Google AI embeddings instead of OpenAI
DEEPWIKI_EMBEDDER_TYPE=google
# Optional: OpenRouter models
OPENROUTER_API_KEY=your_openrouter_api_key
# Optional: Azure OpenAI models
AZURE_OPENAI_API_KEY=your_azure_openai_api_key
AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint
AZURE_OPENAI_VERSION=your_azure_openai_version
# Optional: External Ollama server
OLLAMA_HOST=your_ollama_host

Start the stack

docker-compose up

Docker Compose will build the image (first run only), start the container, and expose the API on port 8001 and the frontend on port 3000. Open http://localhost:3000 once the health check passes.The docker-compose.yml pre-configures the ~/.adalflow volume mount for data persistence and binds ./api/logs so log files survive restarts.

Pull the pre-built image from GitHub Container Registry and run it directly with environment variables:

# Pull the image
docker pull ghcr.io/asyncfuncai/deepwiki-open:latest

# Run the container
docker run -p 8001:8001 -p 3000:3000 \
  -e GOOGLE_API_KEY=your_google_api_key \
  -e OPENAI_API_KEY=your_openai_api_key \
  -e OPENROUTER_API_KEY=your_openrouter_api_key \
  -e OLLAMA_HOST=your_ollama_host \
  -e AZURE_OPENAI_API_KEY=your_azure_openai_api_key \
  -e AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint \
  -e AZURE_OPENAI_VERSION=your_azure_openai_version \
  -v ~/.adalflow:/root/.adalflow \
  ghcr.io/asyncfuncai/deepwiki-open:latest

Only include the -e flags for the providers you actually use. At minimum you need one of GOOGLE_API_KEY or OPENAI_API_KEY.

Using a .env file instead of flags

If you prefer to keep your keys in a file rather than passing them as individual flags, mount your .env file into the container:

docker run -p 8001:8001 -p 3000:3000 \
  -v $(pwd)/.env:/app/.env \
  -v ~/.adalflow:/root/.adalflow \
  ghcr.io/asyncfuncai/deepwiki-open:latest

The container reads /app/.env on startup and exports all non-comment variables automatically.

If you want to build the image from source rather than pulling from the registry:

# Clone the repository
git clone https://github.com/AsyncFuncAI/deepwiki-open.git
cd deepwiki-open

# Build the image
docker build -t deepwiki-open .

# Run the container
docker run -p 8001:8001 -p 3000:3000 \
  -e GOOGLE_API_KEY=your_google_api_key \
  -e OPENAI_API_KEY=your_openai_api_key \
  -e OPENROUTER_API_KEY=your_openrouter_api_key \
  -e AZURE_OPENAI_API_KEY=your_azure_openai_api_key \
  -e AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint \
  -e AZURE_OPENAI_VERSION=your_azure_openai_version \
  -e OLLAMA_HOST=your_ollama_host \
  deepwiki-open

Self-signed certificates

If your environment uses self-signed TLS certificates, include them in the build so the container trusts internal services:

Create a certs/ directory in the project root (or any directory name you choose).
Copy your .crt or .pem certificate files into it.
Build with the optional CUSTOM_CERT_DIR argument:

# Build with the default certs/ directory
docker build .

# Or specify a custom directory name
docker build --build-arg CUSTOM_CERT_DIR=my-custom-certs .

The Dockerfile copies certificates into /usr/local/share/ca-certificates/ and runs update-ca-certificates automatically.

The ~/.adalflow volume

The -v ~/.adalflow:/root/.adalflow mount is what makes your data persist between container runs. DeepWiki writes three categories of data to this path:

Path	Contents
`~/.adalflow/repos/`	Cloned repository source code
`~/.adalflow/databases/`	Embeddings and vector indexes
`~/.adalflow/wikicache/`	Cached generated wiki pages

Without this volume mount the container starts fresh on every run — all previously generated wikis and their embeddings are lost when the container stops.

Embeddings are model-specific. If you switch DEEPWIKI_EMBEDDER_TYPE from openai to google (or to ollama), the existing vectors in ~/.adalflow/databases/ are incompatible with the new model and must be regenerated. Delete the databases directory or re-generate wikis for any repository you want to keep using.

Next steps

Environment variables reference

Full list of all supported environment variables and their defaults.

Use Ollama locally

Run DeepWiki entirely offline with a local Ollama model for generation and embeddings.

Get Started

Configuration

Core Features

Self-Hosting

Reference

Deploy DeepWiki with Docker

Using a .env file instead of flags

Self-signed certificates

The ~/.adalflow volume

Next steps

Environment variables reference

Use Ollama locally

Build docs developers (and LLMs) love

Get Started

Configuration

Core Features

Self-Hosting

Reference

Documentation Index

​Using a .env file instead of flags

​Self-signed certificates

​The ~/.adalflow volume

​Next steps

Environment variables reference

Use Ollama locally

Build docs developers (and LLMs) love

Using a .env file instead of flags

Self-signed certificates

The ~/.adalflow volume

Next steps