Hugging Face introduced Serge, an AI code reviewer for GitHub, on June 12, 2026. The tool is designed to fit into existing workflows without creating new systems for code review. It reviews pull requests using an OpenAI-compatible language model and follows repository-specific rules. Serge can be invoked with a comment such as 'please review,' and it applies the repository's review policy before publishing comments through GitHub's pull request review experience. The tool is open source and available at https://github.com/huggingface/serge. Source: huggingface
Serge allows maintainers to trigger reviews only when needed and lets repositories define what feedback is useful. It can inspect additional repository context and return a review that can be published directly or edited by a human. Maintainers can also edit or discard the model's output before publishing. The tool supports three deployment options: a GitHub Action for quick setup, a GitHub App for organization-wide automation, and a staged web app for human-in-the-loop review. The GitHub Action is the fastest way to start using Serge, requiring only an LLM API key as a repository secret. Source: huggingface
The review policy for a repository is defined in a .ai/review-rules.md file on the default branch. This ensures that pull requests cannot change the policy used to review them. The rules can specify what feedback is useful, such as focusing on correctness, security, and missing tests while avoiding style-only comments or generated files. Repositories can also provide an optional .ai/context-script for extra context or files to skip, and read-only tools rooted in a local checkout. This gives the reviewer bounded access to project context without exposing secrets or arbitrary shell access. Source: huggingface
