1. ai
  2. /tools
  3. /continue-dev

Continue.dev

Continue.dev is an open-source AI coding assistant that runs inside VS Code and JetBrains IDEs — chat, autocomplete, and custom context providers with your choice of models.

Last reviewed: June 2026

Continue config format and hub features change between releases. Verify at continue.dev.

What it is good at

StrengthDetail
Open sourceSelf-host configs; inspect extension code
Multi-IDEVS Code and JetBrains from one config model
Bring your own modelOpenAI, Anthropic, Ollama, Azure, custom endpoints
Context providers@codebase, @docs, @issue, custom providers
config.yamlDeclarative models, rules, and prompts in repo
HubShare configs and blocks (verify license for org use)

Setup

# VS Code: install "Continue" from marketplace
# JetBrains: install Continue plugin from marketplace

# Optional: add config to repo
mkdir -p .continue

Minimal .continue/config.yaml (shape varies by version — verify docs):

models:
  - name: GPT-4o
    provider: openai
    model: gpt-4o
    apiKey: ${{ secrets.OPENAI_API_KEY }}

  - name: Local Llama
    provider: ollama
    model: llama3.1

rules:
  - "Use TypeScript strict. Match existing import style."

Store API keys in Continue's secret storage or env — not committed plaintext.

Modes / core workflow

FeatureUse when
ChatQ&A, explain code, small edits
AutocompleteInline suggestions while typing
EditSelect code → describe change
Agent modeMulti-step edits (if enabled in your version)

Typical loop:

  1. @codebase or highlight files for context
  2. Ask with scoped instruction — file paths, constraints
  3. Accept or reject diff hunks
  4. Run tests locally

Same verification as Copilot — tab/accept only what you understand.

Custom context providers

Continue supports pluggable context — internal docs, Jira, Postgres schema snippets. Useful when @Files is not enough but full MCP is overkill.

Compare with Building MCP Servers for standardized tool access across Cursor, Claude Code, and Continue.

When to use it vs alternatives

Choose ContinueChoose something else
JetBrains + VS Code with one configCursor — deeper agent UX in VS Code fork
Self-hosted / local models (Ollama)Copilot — enterprise at scale
Open-source requirementClaude Code — terminal agent + hooks
Lightweight chat + autocompleteCline — full autonomous agent in VS Code

See Choosing an AI Coding Tool.

Limitations

  • Agent capabilities less mature than Cursor or Cline in some versions
  • Team audit and SSO depend on deployment model (cloud hub vs self-hosted)
  • Config YAML schema evolves — pin docs version when onboarding teams