Gitea AI Webhook Bot

A webhook-based service that automatically generates AI code reviews for Gitea Pull Requests using xai-review.

Features

• Integrated with AI: Uses advanced LLMs for code analysis.
• Zero-config per repository: Works via a single Gitea System Webhook (or per-repo webhook).
• Custom LLM Support: Configuring it to use OpenAI, Ollama, vLLM, or any OpenAI-compatible endpoint.
• Secure: Runs in an isolated Docker container with strict context passed via environment variables.

Deployment

Docker Compose

Run the container using the provided compose.yaml. This will start the service in daemon mode.

docker compose up -d

Configuration (Environment Variables)

Create a .env file with the following variables:

Variable	Required	Description	Default
GITEA_TOKEN	Yes	A Gitea Access Token. Required permissions: repository (Read), issue (Read and Write).
GITEA_API_URL	Yes	Full URL to your Gitea API (e.g., https://git.example.com/api/v1).
PORT	No	Host port to expose the webhook server on.	3000
OPENAI_API_KEY	Yes	API Key for your LLM provider. Use dummy for local models if no auth needed.
OPENAI_BASE_URL	No	Base URL for the LLM API. Set this for Ollama/vLLM (e.g. http://host.docker.internal:11434/v1/).	https://api.openai.com/v1/
LLM_MODEL	No	The model name to request (e.g. gpt-4o, llama3.1).	gpt-4o
LLM_TEMPERATURE	No	The creativity of the model (0.0 - 1.0).	0.2
LLM_PROVIDER	No	The XAI provider type. Use OPENAI for most compatible services.	OPENAI

Example .env (Ollama)

GITEA_TOKEN=da39a3ee5e6b4b0d3255bfef95601890afd80709
OPENAI_API_KEY=dummy
OPENAI_BASE_URL=http://host.docker.internal:11434/v1/
LLM_MODEL=llama3.1

Setup in Gitea

1. Navigate to Site Administration > Webhooks (for all repos) OR Repo Settings > Webhooks.
2. Add a Gitea webhook.
3. Target URL: http://<container-ip>:3000/webhook (e.g., http://gitea-ai-review:3000/webhook).
4. HTTP Method: POST.
5. Trigger On: Custom Events -> Pull Request.