Continue

Core Positioning#

Continue brings the power of Large Language Models (LLMs) into software development's Continuous Integration (CI) and daily coding workflows. Its core mechanism allows developers to write "check logic" in natural language within Markdown files (.continue/checks/), executed by AI agents during Pull Request (PR) stages or in local development environments.

Problems Solved:

• Reduce AI Noise: Define clear check rules to avoid generic AI suggestions interfering with development workflows
• Automated Code Review: Automatically detect security vulnerabilities (e.g., hardcoded keys), code standard violations, and provide actionable fix Diffs
• Continuous Compliance & Quality Control: Integrate AI checks as a version-controlled, traceable "test" step in GitHub Status Checks

Use Cases:

• GitHub PR automated review and security scanning
• Local IDE-assisted programming with real-time checks
• Cloud Agents-based scheduled tasks (Cron) or Webhook-triggered automated fixes (e.g., auto-responding to Sentry alerts)

AI Checks & Definition#

Markdown-Defined Rules: Create .md files in the .continue/checks/ directory, using YAML Frontmatter to define name and description, with the body containing natural language Prompts.

---
name: Security Review
description: Review PR for basic security vulnerabilities
---
Review this PR and check that:
  - No secrets or API keys are hardcoded
  - All new API endpoints have input validation
  - Error responses use the standard error format

Key Features:

• Natural Language Logic: No need to write complex regex or scripts, describe review requirements directly in English
• Source Control: Check rules are versioned with the codebase, ensuring rules evolve with the code

CI/CD Automation & Fixes#

GitHub Status Checks: Checks run automatically when PRs are submitted, with results displayed directly in PR merge check items (green pass / red fail).

Suggestive Diffs: When checks fail, AI generates suggested code modification Diffs that developers can accept or reject with one click.

Cloud Agents: Supports cloud-hosted execution with triggers including:

• Automatic trigger on PR open
• Scheduled dispatch (Cron)
• Webhook event triggers

Runtime Environments & Interaction#

CLI Agents:

• TUI (Terminal User Interface): Interactive terminal interface with step-by-step approval of AI operations
• Headless Mode: Unattended mode suitable for batch processing automation in servers, containers, or CI environments

IDE Extensions:

• VS Code Extension: Integrated real-time assistance in the editor
• JetBrains Plugin: Support for IntelliJ-based IDEs

Ecosystem Integrations#

• GitHub: Deep PR workflow integration
• Slack: Trigger agent tasks via @Continue mentions
• Sentry: Auto-generate fix PRs when errors are detected
• Snyk: Auto-generate fix code when security vulnerabilities are detected

Installation & Deployment#

Requirements: Node.js v20 or higher

Method 1: Npm Global Install

npm i -g @continuedev/cli
cn

Method 2: Shell Script Install (macOS / Linux)

curl -fsSL https://raw.githubusercontent.com/continuedev/continue/main/extensions/cli/scripts/install.sh | bash

Method 3: Windows PowerShell Install

irm https://raw.githubusercontent.com/continuedev/continue/main/extensions/cli/scripts/install.ps1 | iex

Quick Start#

1. Local Run: Execute cn in terminal after installation to launch TUI interface
2. Cloud Access: Visit Mission Control to configure cloud Agents
3. Define Checks: Create .continue/checks/ folder in project root and add Markdown check files