The self-hosted option is only available for CodeRabbit Enterprise customers with 500 user seats or more. Please contact CodeRabbit Sales to learn more about the CodeRabbit Enterprise plan.

​
Create an Azure DevOps User​

  • Username: Set the username to “CodeRabbit” for easier identification (optional).
  • Profile Image: Use the CodeRabbitAI logo for the user image (optional).
The user installing CodeRabbit must have either of the following permissions:
  • Project Administrator
  • Project Collection Administrator
We recommend using the CodeRabbit logo as the profile picture to ensures easy recognition.

​
Add User to Projects​

Add the CodeRabbit user to each project where you want CodeRabbit to post reviews, with rights to post reviews & open PRs.

​
Create a Personal Access Token for CodeRabbit user​

Generate a personal access token for the CodeRabbit user to be added in the .env file as AZURE_DEVOPS_BOT_TOKEN. Necessary Scopes:
  • Code - Full
  • Work Items - Read, write, and manage
  • Build - Read
Consult official CodeRabbitAI documentation for a detailed guide on creating personal access tokens.

​
Add the necessary webhooks to each project​

1

Navigate to project's Service Hooks Page

Go to the service hooks configuration page in the desired Azure DevOps project.
2

Add the following webhooks

Create service hooks for these events:
  • Pull request created
  • Pull request updated
  • Pull request commented on
3

Add Webhook URL

Enter the URL pointing to the CodeRabbit service, followed by /azure_webhooks (e.g., http://127.0.0.1:8080/azure_webhooks) for each webhook.

​
Prepare an .env file​

Create an .env file with the following content:
.env
# if using OpenAI
LLM_PROVIDER=openai
LLM_TIMEOUT=360000
OPENAI_API_KEYS=<openai-key>
OPENAI_BASE_URL=[<openai-base-url>]
OPENAI_ORG_ID=[<openai-org-id>]
OPENAI_PROJECT_ID=[<openai-project-id>]

# if using Azure OpenAI
LLM_PROVIDER=azure-openai
LLM_TIMEOUT=360000
AZURE_OPENAI_ENDPOINT=<azure-openai-endpoint>
AZURE_OPENAI_API_KEY=<key>
# it is recommended to deploy text-embedding-3-large, gpt-4.1-mini, o4-mini, o3, gpt-4.1 (optionally).
AZURE_TEXT_EMBEDDING_3_LARGE_DEPLOYMENT_NAME=<text-embedding-3-large-deployment-name>
AZURE_GPT41MINI_DEPLOYMENT_NAME=<gpt-4.1-mini-deployment-name>
AZURE_O4MINI_DEPLOYMENT_NAME=<o4-mini-deployment-name>
AZURE_O3_DEPLOYMENT_NAME=<o3-deployment-name>
AZURE_GPT41_DEPLOYMENT_NAME=[<gpt-4.1-deployment-name>]
# optionally, deploy gpt-4o-mini instead of gpt-4.1-mini
AZURE_GPT4OMINI_DEPLOYMENT_NAME=[<gpt-4o-mini-deployment-name>]
# optionally, deploy o3-mini instead of o4-mini
AZURE_O3MINI_DEPLOYMENT_NAME=[<o3-mini-deployment-name>]
# optionally, deploy o1 instead of o3
AZURE_O1_DEPLOYMENT_NAME=[<o1-deployment-name>]

# OAuth2 Configuration (optional)
# This will use client_credentials flow to get an access token,
# and use it to make requests to the LLM provider.
# Here is more information on this flow: https://learn.microsoft.com/en-us/entra/identity-platform/v2-oauth2-client-creds-grant-flow#first-case-access-token-request-with-a-shared-secret
# It is expected that the response from the OAuth2 server will be in the format
# {
#   "access_token": "<access-token>",
#   "token_type": "Bearer",
#   "expires_in": 3599,
# }
OAUTH2_ENDPOINT=[<endpoint>]
OAUTH2_CLIENT_ID=[<client-id>]
OAUTH2_CLIENT_SECRET=[<client-secret>]
OAUTH2_SCOPE=[<oauth2-scope>]

HTTP_PROXY=[<http-proxy-url>]
HTTPS_PROXY=[<https-proxy-url>]
NO_PROXY=[<no-proxy-url>]

# if using AWS Bedrock
# it is required to have access to claude-3-haiku, claude-3-5-haiku, claude-sonnet-4, claude-opus-4.
LLM_PROVIDER=bedrock-anthropic
LLM_TIMEOUT=360000
AWS_ACCESS_KEY_ID=<aws-access-key>
AWS_SECRET_ACCESS_KEY=<aws-secret-access-key>
AWS_REGION=<aws-region>
# optionally, use cross-region inference to access models in other regions
# if this is set to `true`, CodeRabbit will access models from `us`, `eu`, or `ap` regions based on the AWS_REGION value.
AWS_USE_CROSS_REGION_INFERENCE=[<true>]

TEMP_PATH=/cache

SELF_HOSTED=azure-devops

AZURE_DEVOPS_BOT_TOKEN=<personal-access-token>
AZURE_DEVOPS_BOT_USERNAME=<bot-user-username>

CODERABBIT_LICENSE_KEY=<license-key>

CODERABBIT_API_KEY=<coderabbitai-api-key>
ENABLE_METRICS=[true]

# CodeRabbit - KNOWLEDGE BASE
#
# Requisite: To enable CodeRabbit's knowledge base, you need to provide an object store URI.
OBJECT_STORE_URI=[<object-store-uri>] # All major object stores are supported e.g., s3://bucket/path/to/database, gs://bucket/path/to/database, etc.

ENABLE_KNOWLEDGE_BASE=[true] # Global feature flag to enable/disable all knowledge base features.

ENABLE_PRS_INDEX=[true] # Feature flag to enable/disable PRs indexing.
ENABLE_ISSUES_INDEX=[true] # Feature flag to enable/disable Issues indexing.
ENABLE_LEARNING_INDEX=[true] # Feature flag to enable/disable Learning indexing.
ENABLE_CODE_GUIDELINES_INDEX=[true] # Feature flag to enable/disable Code Guidelines indexing.
ENABLE_CODE_INDEX=[true] # Feature flag to enable/disable Code indexing.

JIRA_HOST=[<jira-host-url>]
JIRA_PAT=[<jira-personal-access-token>]

LINEAR_PAT=[<linear-personal-access-token>]

ENABLE_WEB_SEARCH=[true]
PERPLEXITY_API_KEY=[<perplexity-api-key>]

YAML_CONFIG=[<escaped-yaml-config>]
  • If you are using Azure OpenAI, verify that the model deployment names are in the .env file.
  • Values marked with [] are not optional to provide.
  • You can generate CODERABBIT_API_KEY from CodeRabbit UI -> Organizations Settings -> API Keys.
  • YAML_CONFIG is an optional configuration file that can be used to customize CodeRabbit’s behavior at the deployment level. It takes the same format as the CodeRabbit YAML configuration file. It requires the entire YAML file to be in an escaped string format, for example, YAML_CONFIG="key1: value1\nkey2: value2". You can use Escape YAML to generate the escaped string.

​
Pull the CodeRabbit Docker image​

Authenticate and pull the Docker image using the provided credentials file: 
cat coderabbit.json | docker login -u _json_key --password-stdin us-docker.pkg.dev
docker pull <docker-registry>/coderabbit-agent:latest

​
Verify the image is up ​

You can query /health endpoint to verify that the coderabbit-agent service is up and running. 
curl 127.0.0.1:8080/health

​
Host the image ​

You can host the image on a server, serverless function, or container environment and expose port 8080. Run the Docker image with the equivalent command on your chosen platform, ensuring you replace the .env file path with the path to your actual .env file: 
docker run --env-file .env --publish 127.0.0.1:8080:8080 <docker-registry>/coderabbit-agent:latest