Continue.dev — AI Coding Tool Guide

Open-source AI code assistant. Local model support, custom slash commands, configuration, and integration with Ollama, LM Studio, and cloud APIs.

October 20, 2026
continuedevopen-sourcelocal-modelsollamavscode

Continue.dev

Continue.dev is the leading open-source AI code assistant. It's a VS Code/JetBrains extension that connects to any model — local (Ollama, LM Studio) or cloud (Anthropic, OpenAI, Gemini).

What Makes Continue.dev Different

  • Fully open-source — MIT license, inspect/modify everything
  • Local models — Use Ollama or LM Studio for offline coding
  • Custom slash commands — Create project-specific AI workflows
  • Model flexibility — Switch between local and cloud models in the same session

Getting Started

Install from continue.dev, then configure ~/.continue/config.yaml:

name: My Config
version: 1.0.0
schema: v1
models:
  - name: Claude Sonnet 4.6
    provider: anthropic
    model: claude-sonnet-4-6
    apiKey: ${{ secrets.ANTHROPIC_API_KEY }}
  - name: Llama 4 (Local)
    provider: ollama
    model: llama4
  - name: Starcoder2
    provider: ollama
    model: starcoder2:3b
    roles:
      - autocomplete

Custom Slash Commands

~/.continue/config.yaml:

prompts:
  - name: review
    description: Code review with project standards
    prompt: |
      Review this code against our standards:
      - TypeScript strict mode, no any types
      - Server components unless interactivity needed
      - Error boundaries on route segments
      Return a list of violations and suggested fixes.
  - name: explain
    description: Explain selected code
    prompt: |
      Explain this code at a senior engineer level. Cover: architecture decisions, patterns used, potential issues.

When to Choose Continue.dev

Continue.dev is the natural pick when the constraint is freedom and cost: it's fully open source, runs on local models, and works with whatever provider you already pay for. It fits:

  • Local-first developers — Code offline with Ollama or LM Studio, with no cloud calls and no per-token bills.
  • Heterogeneous teams — Standardize on one extension while letting each developer pick their own model (local or cloud) rather than forcing a single vendor.
  • Custom workflows — The prompts section in config.yaml lets you bake team-specific review, explain, and scaffold workflows directly into the assistant.

For the fully open-source, terminal-based agent alternative, see OpenCode. If your team lives in GitHub, Copilot offers deeper GitHub integration; for a VS Code fork with agent mode, see Cursor.

Frequently Asked Questions

Sources