What it does
AI-Codereview-Gitlab is a service you deploy yourself to run automated code review with a large language model. When a Merge Request or Push happens in GitLab, a webhook calls the service, which sends the change to the configured model and writes the review back as a note on the Merge Request or commit. The model is selected in an environment file and can come from DeepSeek, ZhipuAI, OpenAI, Anthropic, Tongyi Qianwen or Ollama. Review results can be pushed to DingTalk, WeCom or Feishu, and the tool can also build daily progress reports from commit records in GitLab, GitHub or Gitea. A web dashboard lists review records with project and developer statistics, and the tone of the reviews can be set to professional, sarcastic, gentle or humorous. An optional agentic mode lets the model read files and run read-only shell commands in a local clone of the repository, and it falls back to diff-only review if any stage fails. The project notes that this mode uses more tokens, disk space and time. It can be deployed with Docker Compose or in a local Python 3.10+ environment. The project is licensed under Apache-2.0, and its README also points to a separate Pro edition.
Capabilities
- Automatic review of GitLab Merge Request and Push events through a webhook
- Model choice across DeepSeek, ZhipuAI, OpenAI, Anthropic, Tongyi Qianwen or Ollama
- Review notifications to DingTalk, WeCom or Feishu
- Daily reports built from GitLab, GitHub or Gitea commits
- Dashboard with review records and project and developer statistics
- Optional agentic mode with read-only exploration of a cloned repository
When to use it
- A team on GitLab wants an automated first-pass review on every merge request
- You want to run the review service on your own server and pick your own model provider
- Your team works in DingTalk, WeCom or Feishu and wants review results posted there
