DOCSEARCHbyAlgolia
Submit Docs

Continue Docs: AI Coding Assistant and Automation

website·Indexed Apr 13, 2026

https://docs.continue.dev/

Documentation for Continue, an open-source AI coding assistant. Features include AI checks for pull requests defined in .continue/checks/, a CLI tool (cn) for terminal-based agent tasks, and Mission Control for cloud agent orchestration. Supports integrations with GitHub, Supabase, Sentry, Snyk, Slack, PostHog, Atlassian, and Netlify. Covers configuration, tool permissions, and IDE extensions for VS Code and JetBrains.

Tokens
4.2K
Snippets
6
Records
48
Agent score
50%

What's inside Continue

  1. Overview: Prompts in Continue

    Prompts in Continue are included as user messages to kick off tasks for Agent mode, Plan mode, and Chat mode. They are especially useful for repetitive and/or complex tasks. Prompts become available as slash commands when you set invokable: true in the markdown frontmatter.

  2. When to use @Codebase vs @Folder context providers

    @Codebase works well for high-level questions about your codebase (e.g., 'How do I add a new endpoint?') and generating code using existing patterns as reference. @Folder is better for questions about a specific folder, increasing result relevance. @Codebase is not suitable when you need the LLM to see literally every file (e.g., 'Find all usages of function foo') or for refactoring tasks that require comprehensive codebase scanning.

  3. Development Data Collection Overview

    Continue automatically collects development data as a byproduct of LLM-aided development. By default, this data is saved to .continue/dev_data on your local machine. This data captures how you build software and can be useful for understanding your development patterns.

  4. Metrics: Agent observability for Cloud Agents

    Metrics provides operational observability for Cloud Agents and automated workflows. It answers the question: What are my AI agents doing, and is it working? Use Metrics to monitor agent activity, understand human intervention, measure success rates, and evaluate the cost and impact of AI-driven work across repositories.

  5. Telemetry collection overview

    Continue open-source extensions collect anonymous usage data to improve the product. All data is anonymized and stripped of PII before being sent to PostHog. Collection can be disabled at any time.

  6. What Metrics track about Cloud Agents

    Agent observability focuses on four key areas: (1) Runs and execution frequency, (2) Success vs. human intervention rates, (3) Pull request outcomes, and (4) Cost per run and per workflow. This replaces traditional observability metrics like logs and latency for AI agent contexts.

  7. Model Capabilities Overview

    Model capabilities tell Continue what features a model supports: tool_use enables tools and function calling (required for Agent mode), and image_input enables image processing. Without proper configuration, you may encounter issues like Agent mode being unavailable, tools not working, or image uploads being disabled.

  8. MCP servers overview and capabilities

    Model Context Protocol (MCP) servers enable Continue to connect to external tools, systems, and databases. They provide integration with external tools and systems, create extensible interfaces for custom capabilities, support complex interactions with your development environment, allow partners to contribute specialized functionality, and connect to databases to understand schema and data models during development.

  9. Prompts Overview

    Prompts are specialized instructions that shape how Continue models and agents respond to specific tasks. They define interaction patterns for particular frameworks, encode domain expertise for specific technologies, ensure consistent guidance aligned with organizational practices, can be shared and reused across multiple agents, and act as automated code reviewers that enforce consistency across teams.

  10. Query the Continue Documentation MCP Server

    After setup, ask natural language questions about Continue features directly in conversations. Useful query types include: model configuration (e.g., adding models from providers like Bedrock), available context providers, and customization rules. The server requires internet access to search documentation.

  11. Use Chat mode for interactive code assistance

    Highlight code in your editor, use keyboard shortcuts to add it to Chat, then ask questions about the code. Try prompts like 'What sorting algorithm is this and how can I optimize it?', 'Show me how to implement quicksort instead', 'What's the time complexity?', or 'Can you write unit tests for this function?' Use Chat for code reviews, debugging help, learning concepts, or brainstorming.

  12. Create slash command prompts with invokable frontmatter

    Set invokable: true in the YAML frontmatter of a markdown file to make it a slash command. The name field defines the command name and description provides context. The prompt can be combined with other instructions and highlighted code for additional context in Chat, Plan, and Agent modes.

    Example frontmatter:

    ---
name: Explain invokable
description: Explains what happens when you set invokable to true
invokable: true
---