Continue

At a Glance:

Continue is an open-source CLI tool that runs source-controlled AI agents as enforceable GitHub status checks on pull requests, with each agent defined in a repo's .continue/checks/ markdown files.

Overview:

Continue is a developer tool that brings AI-powered code review into CI pipelines by running agents as GitHub status checks on every pull request. Instead of relying on external review services, teams can define custom AI checks as markdown files stored directly in the repository under .continue/checks/. Each check passes green or blocks with a red status and a suggested diff. The tool is installed as a command-line interface (cn) and does not require a separate web application or VS Code extension for the CI enforcement workflow, though a VS Code extension exists separately.

Key Decision Points:

• Source-controlled checks: AI agents are defined as markdown files inside the repository, meaning checks are versioned and reviewed alongside code.

• CI-native enforcement: Checks run as GitHub status checks, turning AI review suggestions into enforceable merge requirements rather than advisory comments.

• CLI-based execution: The tool provides a command-line interface (cn) for installation and execution, designed for CI environments.

• Diff-based feedback: When a check fails, the agent returns a suggested diff, giving developers actionable code changes rather than just problem descriptions.

• Installation scope: Supports macOS, Linux, and Windows via package managers or npm with Node.js 20+.

Core Features:

• Markdown-defined AI agents: Each check is a markdown file that defines agent behavior and lives in the repo at .continue/checks/.

• GitHub status check integration: Agents run as native GitHub status checks on pull requests, blocking merges when issues are found.

• Suggested diffs on failure: Failing checks return a red status with a concrete diff, providing immediate fix proposals.

• Cross-platform CLI: The cn CLI installs on macOS, Linux, and Windows through package managers or npm.

Use Cases:

• Automatic security review: Run a security-focused AI agent on every pull request as a required status check.

• Code quality enforcement in CI: Make AI code review a blocking step in the merge pipeline instead of an optional review tool.

• Custom review rules as code: Define and version-control project-specific review standards alongside the codebase.

Open-Source Alternative Value:

Continue provides an open-source approach to AI-powered pull request review that keeps both the review logic and the execution tooling transparent. Checks are stored as plain markdown files within the repository, meaning teams can inspect, modify, and version them using standard git workflows. The CLI is open-source and runs in any CI environment capable of executing status checks, without depending on a hosted SaaS review service.