continuedev/continue

Continue is a tool that brings automated AI code review directly into your software development workflow, specifically by running checks on every pull request as part of your continuous integration pipeline. The core idea is that you write review criteria in simple text files stored in your repository, and Continue uses AI agents to evaluate whether new code meets those criteria, flagging issues automatically without a human reviewer needing to look at every pull request first. The way it works is straightforward: you create markdown files in a folder called .continue/checks/ inside your project. Each file describes a check, such as "make sure no API keys are hardcoded" or "confirm all new endpoints validate input." When a pull request is opened, Continue runs these checks as GitHub status checks, the same kind of green or red indicators you already see from automated tests. If a check fails, Continue not only marks it red but also suggests a concrete code diff to fix the problem. This is useful for development teams who want consistent code quality standards enforced automatically. Instead of relying on human reviewers to catch the same recurring issues, security antipatterns, missing error handling, style violations, you encode those rules once and the AI enforces them on every future change. Because the checks are just text files in version control, the whole team can contribute to and audit them like any other code. The project includes a command-line tool called the Continue CLI, installable via a shell script or npm, as well as extensions for code editors including VS Code and JetBrains IDEs. The stack is TypeScript, and the system integrates with large language models to power the actual review logic.