Overview:
Continue is an open-source CLI tool that runs AI agents as GitHub status checks on every pull request. It enables teams to enforce automated code reviews directly from the repository by defining agents as markdown files at .continue/checks/. Each agent can validate code against custom rules, such as security or style, and either pass the check or fail with a suggested diff. This project is designed for developers and engineering teams who want to integrate source-controlled, customizable AI checks into their CI pipeline.
Core Features:
Source-controlled checks: Agents are defined as markdown files in the repository under
.continue/checks/, making them versioned and auditable alongside the codebase.PR status checks: Runs agents on every pull request as GitHub status checks, providing clear pass/fail results.
Suggested diffs: Failing checks include a suggested diff, helping reviewers and authors quickly understand the required changes.
Use Cases:
Automated security review on PRs: Teams can define a security-focused agent that scans pull requests for common vulnerabilities, either passing or suggesting fixes.
Custom code style enforcement: Developers can create agents that enforce team-specific coding conventions or style rules within the CI pipeline.
Configurable CI gatekeeping: Engineering teams can use agents as automated gatekeepers for any custom policy or rule applied to new code changes.
Why It Matters:
Continue brings AI code review into the CI workflow without external proprietary services. By keeping check definitions as markdown files in the repository, teams maintain full transparency over what rules are enforced and how agents are configured. The open-source CLI ensures that the review process is self-hosted and directly integrated with GitHub, giving teams control over their code validation pipeline.
