Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/AsyncFuncAI/deepwiki-open/llms.txt

Use this file to discover all available pages before exploring further.

DeepWiki Open is configured entirely through environment variables and JSON config files. Set these variables in a .env file at the project root or pass them directly to your Docker container. All variables are optional unless explicitly marked as required for a specific feature.

​
API keys

Each model provider and embedding provider requires its own API key. You only need the keys for the providers you intend to use.
​
GOOGLE_API_KEY
string
API key for Google Gemini models and Google AI embeddings. Required when using the google provider for text generation or when DEEPWIKI_EMBEDDER_TYPE=google. Obtain from Google AI Studio.
​
OPENAI_API_KEY
string
API key for OpenAI models and embeddings. Required when using the openai provider for text generation or when DEEPWIKI_EMBEDDER_TYPE=openai (the default embedder). Also required when using OpenAI-compatible embedding endpoints via OPENAI_BASE_URL. Obtain from OpenAI Platform.
​
OPENROUTER_API_KEY
string
API key for OpenRouter, which provides access to models from OpenAI, Anthropic, Google, Meta, Mistral, and more through a single API. Required only when using the openrouter provider. Obtain from openrouter.ai.
​
AZURE_OPENAI_API_KEY
string
API key for your Azure OpenAI resource. Required when using the azure provider. Must be combined with AZURE_OPENAI_ENDPOINT and AZURE_OPENAI_VERSION. Obtain from the Azure Portal.
​
AZURE_OPENAI_ENDPOINT
string
The endpoint URL for your Azure OpenAI resource (e.g. https://my-resource.openai.azure.com/). Required when using the azure provider.
​
AZURE_OPENAI_VERSION
string
The Azure OpenAI API version string (e.g. 2024-02-01). Required when using the azure provider.
​
AWS_ACCESS_KEY_ID
string
AWS access key ID for authenticating with Amazon Bedrock. Required for Bedrock unless you are using instance-level or role-based IAM credentials.
​
AWS_SECRET_ACCESS_KEY
string
AWS secret access key paired with AWS_ACCESS_KEY_ID. Required for Bedrock when not using instance or role-based credentials.
​
AWS_SESSION_TOKEN
string
AWS session token for temporary STS credentials. Required only when using short-lived credentials obtained via AssumeRole or similar mechanisms.
​
AWS_REGION
string
AWS region where Bedrock is accessed. Defaults to us-east-1 when not set.
​
AWS_ROLE_ARN
string
ARN of an IAM role to assume before calling Bedrock. When set, the Bedrock client will call STS AssumeRole automatically. Leave unset if your credentials already have the necessary permissions.
​
OLLAMA_HOST
string
Base URL of the Ollama server. Defaults to http://localhost:11434. Set this only when your Ollama instance runs on a remote host or a non-default port.

​
Embedder configuration

​
DEEPWIKI_EMBEDDER_TYPE
string
Selects the embedding provider used to create vector representations of repository code. Accepted values are openai, google, ollama, and bedrock. Defaults to openai.Changing this value after a repository has already been indexed requires re-generating the embeddings, because different models produce incompatible vector spaces.
​
OPENAI_BASE_URL
string
Overrides the base URL for OpenAI API calls. Use this to point the OpenAI client at a private or enterprise endpoint, or at any OpenAI-compatible third-party service (such as Alibaba Qwen). When set, the same key is also used by the embedder if DEEPWIKI_EMBEDDER_TYPE=openai.Example: https://custom-api-endpoint.com/v1

​
Server configuration

​
PORT
string
Port that the API server listens on. Defaults to 8001. If you change this, update SERVER_BASE_URL to match so the frontend can reach the backend.
​
SERVER_BASE_URL
string
Full base URL the frontend uses to reach the API. Defaults to http://localhost:8001. Update this whenever you change PORT or deploy the API to a remote host.

​
Authorization

​
DEEPWIKI_AUTH_MODE
string
Enables authorization mode when set to true or 1. In this mode the frontend displays an input field for a secret code before allowing wiki generation. Defaults to false.
​
DEEPWIKI_AUTH_CODE
string
The secret code users must supply when DEEPWIKI_AUTH_MODE is enabled. Has no effect when auth mode is disabled. Only used if DEEPWIKI_AUTH_MODE is true or 1.

​
Logging

​
LOG_LEVEL
string
Controls the verbosity of application logs. Accepted values are DEBUG, INFO, WARNING, ERROR, and CRITICAL. Defaults to INFO.
​
LOG_FILE_PATH
string
Path where log output is written. Defaults to api/logs/application.log. The application enforces that this path resides within the project’s api/logs directory to prevent path traversal. When running via Docker Compose, api/logs is bind-mounted to ./api/logs on the host so logs persist across container restarts.

​
Configuration directory

​
DEEPWIKI_CONFIG_DIR
string
Path to a directory containing custom generator.json, embedder.json, and repo.json config files. When not set, DeepWiki looks for these files in the api/config/ directory bundled with the source code. Set this when you want to supply your own model or embedder configuration without modifying the source.

​
Minimal .env examples

The following snippets show the smallest viable configuration for each common setup. 
# Google Gemini + Google embeddings
GOOGLE_API_KEY=your_google_api_key
DEEPWIKI_EMBEDDER_TYPE=google

# OpenAI models + OpenAI embeddings (default embedder)
OPENAI_API_KEY=your_openai_api_key

# Azure OpenAI
AZURE_OPENAI_API_KEY=your_azure_key
AZURE_OPENAI_ENDPOINT=https://my-resource.openai.azure.com/
AZURE_OPENAI_VERSION=2024-02-01
OPENAI_API_KEY=your_openai_api_key   # still needed for default embedder

# Local Ollama (no API key required)
DEEPWIKI_EMBEDDER_TYPE=ollama
OLLAMA_HOST=http://localhost:11434

# AWS Bedrock
AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AWS_REGION=us-east-1
DEEPWIKI_EMBEDDER_TYPE=bedrock

Build docs developers (and LLMs) love

Get started for freeTalk to us