Continue Alternatives: Where to Go Now the Repo Is Read-Only

Continue's GitHub repository says it "is no longer actively maintained and is read-only," after a final 2.0.0 release in June 2026. Alternatives include Cline and Kilo Code for agents in VS Code and JetBrains, Tabby and twinny for local completion, the Claude Code, Codex and Copilot extensions, and OpenCode or Aider in the terminal.

Safeguard is an alternative you can switch to without losing what you already pay for. Bring the Claude, ChatGPT or Codex plan you already pay for and use it inside the Safeguard CLI. Safeguard adds no extra charge for that usage; your plan's own usage limits and terms still apply. You can also connect Claude or ChatGPT to the Safeguard MCP server and use Safeguard from the chat you already use. In the editor, Safeguard's IDE extensions for VS Code, JetBrains and Open VSX editors add inline security scanning, guardrail enforcement and one-click remediation.

Every fact here comes from a project's own repository, licence file or docs, read on 21 October 2026. There are no rankings.

What changed with Continue, and what still works

Continue is an Apache 2.0 "coding agent available as a CLI, VS Code extension, and JetBrains plugin." Its docs describe four modes in the editor: Agent, Chat, Edit and Autocomplete. The CLI, cn, handles terminal work, and cn -p "your prompt" runs one headless turn for scripts and CI.

The README now opens with a note: "The continuedev/continue repository is no longer actively maintained and is read-only for all users." The team describes a final 2.0.0 release of the VS Code extension, CLI and JetBrains plugin, which removed anonymous telemetry, pulled out authentication and fixed bugs. GitHub's release list shows v2.0.0 for VS Code published on 19 June 2026, marked Latest.

What that means in practice:

• It still installs. The VS Code extension remains on the Marketplace and Open VSX, the CLI on npm, and the JetBrains plugin through GitHub Releases. The README recommends the CLI over the JetBrains plugin.
• The code is still Apache 2.0. You can fork it and maintain your own build. The README says the team hopes the codebase "continues to serve as a foundation for others."
• No upstream fixes. Issues and pull requests on a read-only repo go nowhere. New model releases, editor API changes and dependency updates become your problem.

If you need to stay on Continue for a while, treat it like any frozen dependency. Note the version you run, keep a copy of the installer, and decide who owns a fork if something breaks. Then plan the move.

So the question for most teams is which part of Continue they relied on. Continue's config.yaml assigns models to roles: chat, edit, apply, autocomplete, embed and rerank. Teams usually leaned on one or two of those. Pick a replacement per role rather than looking for one tool that matches everything.

Continue alternatives at a glance

What each tool replaces

Models and approvals

Notes on each tool

Cline. Cline is an Apache 2.0 agent with Plan and Act modes, and you can set a different model for each. In Act mode "every file edit and terminal command requires your approval" unless you enable auto-approve, and checkpoints in a shadow Git repo let you roll back. Before adopting it for JetBrains, note the repo says "we are not open-sourcing JetBrains plugins."

Kilo Code. Kilo describes itself as "the open source coding agent for building with AI in VS Code, JetBrains, or the CLI," under MIT. Its CLI "is a fork of OpenCode." It is the closest match to Continue's surface list, since it covers the same three. Its cloud agent and PR reviews run on Kilo's platform, so decide whether you want those.

Tabby. Tabby is "a self-hosted AI coding assistant," with no database or cloud service needed and support for consumer-grade GPUs. You run the server, then connect the editor with Tabby: Connect to Server..., an endpoint URL and an account token. Check the licence split: code outside ee/ is Apache 2.0, and ee/ has its own licence. The latest tagged release is v0.32.0 from January 2026.

twinny. twinny covers completion, inline edit, chat, code review and an experimental agent mode in VS Code, against a model server you choose, with "no telemetry, no sign-in." It is the closest match to Continue's local autocomplete setup. It is maintained by one person, and its team gateway needs a paid licence for more than five developers or for its policy, recording and plugin features.

GitHub Copilot in VS Code. Copilot's chat and agents now live in the VS Code repository. You need a Copilot plan, and Copilot Free exists. Admins can enforce manual approvals for specific tools and require terminal sandboxing through managed settings.

Claude Code. The VS Code extension needs VS Code 1.94 or later, and the JetBrains plugin runs the claude CLI in the IDE terminal, so you install both. Sign in with a paid Claude subscription (Pro, Max, Team or Enterprise) or a Claude Console account. Claude Code reads CLAUDE.md and can read a repository's AGENTS.md instead.

Codex. OpenAI's extension runs Codex in VS Code and compatible editors and can hand longer tasks to Codex Cloud. Codex is included in ChatGPT Free, Go, Plus, Pro, Business, Edu and Enterprise, and an API key also works for the CLI, SDK and extension. Codex reads AGENTS.md from your Codex home and from the project root down to the current folder.

OpenCode. OpenCode is "the open source coding agent," MIT licensed, with 75+ providers and local models. Run opencode in VS Code's integrated terminal and its extension installs. Most permissions default to allow, so set edit and command permissions to ask before real work.

Aider. Aider is "AI pair programming in your terminal," Apache 2.0. With --watch-files, a comment ending in AI! in any editor asks it to make a change and AI? asks a question. Its latest tagged release is v0.86.0 from August 2025.

Migrating your Continue setup step by step

Continue kept most of its behaviour in two places: config.yaml and .continue/rules. Here is how to carry both across.

1. Export what you have. Copy config.yaml and the .continue/rules folder somewhere outside the repo. Note each model's provider, model and roles, every mcpServers entry, and any prompts you invoke with a slash.

A trimmed Continue config often looks like this, with one model per job:

name: team-assistant
version: 1.0.0
schema: v1
models:
  - name: Local completion
    provider: ollama
    model: qwen2.5-coder:1.5b
    roles:
      - autocomplete
  - name: Chat
    provider: anthropic
    model: your-chat-model
    roles:
      - chat
      - edit
      - apply
mcpServers:
  - name: internal-docs
    command: npx
    args: ["-y", "your-mcp-server"]

Each block answers a migration question. The autocomplete model needs a completion tool. The chat, edit and apply model needs an agent. Each mcpServers entry needs re-adding by hand.

2. Map each role to a tool. A typical split looks like this:

3. Move your rules. Continue rules are Markdown files with frontmatter in .continue/rules, applied to Agent, Chat and Edit but not autocomplete. Cline reads .clinerules/, .cline/rules/ and AGENTS.md. Claude Code reads CLAUDE.md or AGENTS.md. Codex reads AGENTS.md. Writing one AGENTS.md at the repo root covers three of them.

A Continue rule converts with little more than a copy. This:

---
name: Dependency rule
alwaysApply: true
---

- Pin exact versions when adding a package.
- Never add a package without saying why in the PR description.

becomes a section of AGENTS.md:

## Dependencies

- Pin exact versions when adding a package.
- Never add a package without saying why in the PR description.

Frontmatter such as globs has no direct equivalent in AGENTS.md. Codex reads nested AGENTS.md files from the project root down to the current folder, so a rule for one part of the tree can live in an AGENTS.md in that subfolder and applies when Codex works from there.

4. Re-add MCP servers one at a time. Check which tools each server exposes in the new agent. Continue's agent defaulted each tool to Ask First; do not copy an "Automatic" setting into a tool whose defaults are already permissive.

5. Match Continue's approval baseline. Set Cline to Act with auto-approve off, Claude Code to Manual, VS Code agents to Manual permissions, OpenCode's edit and command permissions to ask, and the Safeguard CLI to Ask.

6. Point local models at the new tool. If you ran Ollama or LM Studio for Continue, twinny finds them on their usual ports at first start, and Cline, OpenCode and Aider list them as providers. Tabby serves its own models.

7. Baseline your dependencies before the first session. Scan the repo, save the result, and compare after the agent's first task. The commands are in the next section.

Do the work properly with the Safeguard IDE extensions and CLI

Continue users had one tool in the editor and one in the terminal. Safeguard covers both, and the assistant knows your vulnerabilities. You keep the Claude, ChatGPT or Codex you already pay for.

In the editor

The download page lists Safeguard's IDE extensions for "inline security scanning, guardrail enforcement, and one-click remediation inside your editor": version 5.1.0 for VS Code and VS Code Insiders on the Marketplace, version 5.1.0 on Open VSX for Cursor, Windsurf, Codium and compatible editors, and version 5.0.1 for JetBrains IDEs including IntelliJ IDEA, WebStorm, PyCharm and GoLand. Like Continue, it covers both editor families.

Install the CLI

The Safeguard CLI scans code, container images and SBOMs on your own machine, sends only the results to the platform, and includes an assistant that can fix what it finds. It writes code, not only scans.

macOS and Linux:

curl -fsSL https://cli.safeguard.sh/install | bash

Windows:

irm https://cli.safeguard.sh/install.ps1 | iex

It is a single static binary with no Node, Python or JVM needed, for macOS, Linux and Windows on x86_64 or arm64 (Windows is x86_64 only). The download page lists version 1.4.0.

First session

1. Open a terminal in your project and run safeguard.
2. Type /login to sign in through the browser.
3. Type /scan . to scan the current folder.
4. Ask in plain language, for example "how many critical vulns do we have".
5. Type @ to attach a file from the folder you started in. /resume picks up the last session in that folder.

Writing and fixing code

The assistant has tools for your local files and for your Safeguard data, and picks whichever the question needs. Ask "fix the lodash CVE in this repo" and it edits the code with your approval. It also works as a general terminal assistant: find and read files, edit code and run commands, with your approval, plus everything security. The docs give "what's the latest file in ~/Downloads" as one example.

Use the plan you already pay for, or pick a model

Bring the Claude, ChatGPT or Codex plan you already pay for and use it inside the Safeguard CLI. Safeguard adds no extra charge for that usage; your plan's own usage limits and terms still apply. /model shows or switches the model; see the CLI docs for connecting one.

Continue users are used to choosing a model per job. This demo shows choosing the model for each conversation: Zero z20 answers with no language model at all by turning the question into a platform query, Griffin, Eagle and Lion are Safeguard's models, a tenant admin chooses which model families people may use, and you can bring your own model on your own provider account and route jobs such as chat or summaries to it.

Click through it at your own pace, or open it full screen.

You can also stay in ChatGPT. This demo connects ChatGPT to the Safeguard MCP server with OAuth and chooses which tools ChatGPT can use.

Click through it at your own pace, or open it full screen.

Approvals

Tools that change things, such as writing a file, editing code or running a command, ask before they act. /permissions sets the mode:

• Ask prompts before every action, the closest match to Continue's Ask First.
• Auto approves known actions and asks for anything new. This is the default.
• Yolo approves everything.

Scan before and after any agent

Whichever replacement you pick, check what it changed:

safeguard scan --dir .
cp output/vuln.json vuln-before.json
# run your Cline, Kilo Code, Claude Code, Codex or Copilot session here
safeguard scan --dir .
diff vuln-before.json output/vuln.json

The same command takes --repo <url>, --image <name> or --sbom <file>. Results land under ./output as JSON, including vuln.json and license.json. safeguard validate sbom.cdx.json checks CycloneDX, SPDX and VEX documents. When the local runner handles platform scan jobs, your source code never leaves the machine; only SBOMs and findings are uploaded.

Replace cn -p in CI with safeguard exec

If you used cn -p in a pipeline, the closest Safeguard step is one non-interactive assistant turn:

safeguard exec "summarise the critical findings in this repository"

Pass or fail decisions belong to policies and gates, evaluated in the Safeguard platform, not by the CLI. The platform also offers reachability analysis and autonomous remediation with fix pull requests, off by default. For more on day-to-day use, read the Safeguard CLI as an AI assistant on your machine.

Every surface, one platform

Start in the terminal, continue in the desktop app (3.2.2 for Windows, macOS and Linux) to ask your supply chain anything in plain English, and check findings on your phone with the iOS or Android app. With nothing to install, there is the Cloud CLI on Safeguard's infrastructure, a cloud browser, and a cloud sandbox tenant against a seeded demo repo, free for seven days with no card.

Common mistakes when leaving Continue

Waiting for a fix that will not come. The repository is read-only. If a model or editor update breaks your setup, plan to fork or move.

Looking for one tool that does every role. Continue bundled autocomplete, chat, edit and an agent. Splitting them, for example twinny for completion and an agent for the rest, is often simpler.

Copying permissive tool settings. Continue's agent defaulted to Ask First. OpenCode defaults most permissions to allow, and Claude Code's extension starts in Auto from v2.1.283. Set approvals on day one.

Forgetting rules files. Your .continue/rules will not load anywhere else. Convert them to AGENTS.md or the new tool's format.

Plugging a subscription into a tool the provider has not approved. OpenCode's providers page says plugins exist to use Claude Pro or Max with OpenCode and that "Anthropic explicitly prohibits this." Read your provider's terms first.

Skipping the dependency check. Any agent can add a package version with a known vulnerability. Scan before and after.

FAQ

Is Continue shut down?

The repository says it is "no longer actively maintained and is read-only for all users." The final 2.0.0 release remains installable, and the code stays Apache 2.0, so you can fork it.

What is the closest open-source replacement for Continue?

By surfaces, Kilo Code covers VS Code, JetBrains and a CLI under MIT. Cline covers the same surfaces under Apache 2.0, with a JetBrains plugin that is not open source. For local autocomplete, twinny (MIT) and Tabby (Apache 2.0 outside ee/) are the closest.

Which alternatives work fully with local models?

From their docs: Tabby serves its own models, twinny finds Ollama, LM Studio and llama.cpp, and Cline, OpenCode and Aider list local providers.

Can I switch to Safeguard and keep my Claude, ChatGPT or Codex?

Yes. Bring the Claude, ChatGPT or Codex plan you already pay for and use it inside the Safeguard CLI. Safeguard adds no extra charge for that usage; your plan's own usage limits and terms still apply. You can also connect Claude or ChatGPT to the Safeguard MCP server and use Safeguard from that chat.

Does Safeguard have a JetBrains plugin like Continue did?

Yes. The download page lists the JetBrains plugin at version 5.0.1, alongside VS Code and Open VSX extensions at 5.1.0.

What replaces cn -p in CI?

safeguard exec runs one non-interactive assistant turn. Kilo's kilo run and other headless CLIs also run single tasks. Gates and policies are evaluated in the Safeguard platform.

Does the Safeguard CLI replace Continue feature for feature?

No. It is a terminal assistant that writes and fixes code with your approval and knows your vulnerability data, paired with IDE extensions for scanning and remediation. Use it as your agent, or keep another and scan what it changed.

Try it yourself

This demo sets up the runner on your own machine: pick the platform, start it, approve it with its device code, then use the same CLI to scan from the terminal.

Click through it at your own pace, or open it full screen.

Install from safeguard.sh/download and read the CLI docs. Comparing vendor agents too? See our guides to OpenAI Codex alternatives and the Safeguard CLI as a security-aware coding assistant.

Quick check

Sources

• Safeguard CLI documentation
• Safeguard CLI interactive terminal
• Safeguard download page
• Continue repository
• Continue releases
• Continue docs
• Continue CLI quickstart
• Continue config.yaml reference
• Continue model providers
• Continue rules
• Continue agent tool policies
• Cline repository
• Cline Plan and Act
• Cline rules
• Kilo Code repository
• Tabby repository
• Tabby licence
• Tabby Docker installation
• Tabby VS Code extension
• twinny repository
• Copilot Chat extension repository
• VS Code approvals and permissions
• Claude Code in VS Code
• Claude Code in JetBrains IDEs
• Claude Code memory and CLAUDE.md
• Claude Code licence file
• Codex IDE extension
• Codex agent approvals and security
• Codex pricing and plans
• Codex AGENTS.md
• OpenCode IDE integration
• OpenCode providers
• OpenCode permissions
• Aider repository
• Aider watch mode

Last reviewed: 21 October 2026.