Disclosure: this article contains affiliate links. If you buy through them we may earn a commission at no additional cost to you. Our recommendations are based on the evidence recorded alongside this article, and prices were verified on the date shown.

If you rely on Continue for AI-assisted coding, the project stopped shipping: its GitHub repository is now read-only, and continue.dev says the team joined Cursor. This guide is for developers who need a replacement, not a eulogy — and for anyone who still needs to run local models while they migrate.

What changed and why it matters

The continuedev/continue repository on GitHub is no longer actively maintained. The project's own notice marks it read-only for all users, a status confirmed in a Hacker News discussion of the shutdown, titled plainly: "Continue coding agent is dead. Alternatives?" Read-only means no new commits, no merged pull requests, no patched bugs — the code that exists today on your machine is the code you will have from now on, frozen at whatever version you last pulled.

The same discussion carries a statement from continue.dev: Continue has joined Cursor. That statement explains why the repository froze rather than simply going quiet. It doesn't tell you whether Cursor will fold Continue's specific features into its own product, on what timeline, or whether your existing Continue configuration will have any equivalent there — no such roadmap is part of the confirmed record. What is confirmed is simpler and more actionable than a roadmap: if you have Continue wired into your editor today, you are running unmaintained software, and the decision about what replaces it is now yours, not the project's.

This matters beyond one extension. Continue was, for a meaningful slice of its users, the tool that let them keep coding assistance local — running against self-hosted models rather than a vendor's API. The Hacker News thread that surfaced the shutdown was started by a user (roscas) describing exactly that setup: Continue in VS Code, talking to an offline Ollama instance. Losing an actively maintained option in that specific niche is a different, and arguably larger, problem than losing one more hosted-AI extension among many.

What to look for

Rank your replacement against what actually broke for Continue users, not against a marketing features list.

• Maintenance status first. Check the repository's commit history and issue tracker before you install anything. A tool with an open, active commit log this month beats one with a longer feature list and a stale one. Continue's own history shows how quickly "actively developed" can become "read-only" — verify current state, not reputation.
• A stated commitment, not a rumor. Continue's shutdown came with a clear, on-the-record statement from the vendor about what happened and where the team went. Before you adopt a replacement, look for that same kind of direct statement about its status and direction — not a changelog that quietly stopped, which is exactly how many Continue users first noticed something was wrong.
• Local and offline model support. If your workflow depends on models you run yourself rather than a hosted API, confirm your next tool supports that configuration before you migrate, not after. This was the exact setup described in the Hacker News thread on Continue's shutdown, and it's easy to overlook when a tool's marketing focuses on its hosted, cloud-model experience instead.
• Editor integration you already use. Confirm the extension works in your actual editor — VS Code, JetBrains, or otherwise — rather than assuming parity across editors because a landing page lists all of them.
• A migration path for your config. Custom prompts, rules files, model routing, and context settings are the expensive part to rebuild by hand. Prefer tools that can import your existing setup, or that at least document the format clearly enough that you can port it yourself in an afternoon rather than a week.
• Where the inference actually runs. If you're moving away from a self-hosted model setup, decide up front whether you're staying self-hosted or switching to a vendor's cloud. That's a cost and control decision, not just a technical one, and it's easy to make by default instead of on purpose.

Recommended options

Two realistic paths exist for a former Continue user, and they suit different situations. There's no third path in the record here — no verified pricing or feature comparison exists for other extensions in the evidence available, so this section sticks to what's confirmed rather than guessing at numbers or naming tools this piece can't back with a source.

Move to Cursor. Best for: developers who'd rather follow the team's stated direction than rebuild a workflow from scratch on unfamiliar tooling. Continue.dev's own statement that the project joined Cursor is the clearest signal available about where ongoing development effort is actually going. Skip it if: your hard requirement is a freely modifiable, self-hosted extension you can audit and patch yourself — that's a different category of tool than what's confirmed here, and this piece doesn't have verified specs to compare Cursor against that requirement point by point.

Keep your last working Continue build. Best for: teams already running Continue against a local Ollama instance, per the workflow described in the Hacker News thread, who need continuity while they evaluate alternatives on their own timeline rather than under deadline pressure. Nothing stops a locally installed extension from continuing to function after its repository goes read-only; it simply stops improving. Skip it if: you need active maintenance for any reason — a read-only repository will not receive fixes for bugs, model API changes, or security issues, so this is a bridge, not a destination.

Neither option is a permanent fix by itself. The read-only repository confirms the old path is closed; nothing in the record confirms a specific new path is fully built out yet for every use case Continue covered. Treat both options as what you do this week, not what you commit to for the year, and revisit the decision once you have more information about how Cursor's roadmap actually develops.

Common mistakes

Assuming "read-only" means "still safe to build on." It doesn't. Read-only means the maintainers have stopped accepting changes, which is a different thing from the software still technically working today. Plan your exit on that basis — a frozen dependency, not a healthy one — not on how well it currently runs on your machine.

Not checking whether your configuration survives a move. If you've invested time in custom rules, prompts, or model routing inside Continue, confirm your next tool can import or at least closely replicate that setup before you switch, not after you've already lost a day rebuilding it from memory. Export whatever configuration files you can find first, even if you don't have a destination for them yet.

Ignoring the local-model half of the workflow. The Hacker News thread that surfaced this shutdown was started by someone running Continue against an offline Ollama backend in VS Code, not a hosted model API. If that's your setup too, migrating your editor extension is only half the job; you still need somewhere to run the model itself, reliably, outside of your own laptop if you want it available to a team. If your workflow depends on hosting that inference backend yourself rather than through a vendor's cloud, See current pricing at the merchant so you know what the compute will actually cost before you commit to a self-hosted setup.

Waiting for an official migration guide before acting. Nothing in the record here confirms a published, step-by-step migration path from Continue to any specific replacement. Waiting for one to appear is not a plan — checking your own config, editor, and model requirements against the criteria above is something you can start today, with or without an official guide.

Treating an acquisition announcement as a feature announcement. Continue.dev's statement that the project joined Cursor tells you about a change of ownership and direction, not about specific new capabilities on a specific date. Don't let optimism about "presumably it'll be even better" substitute for checking what's actually shipped before you rely on it.

Make the call now

The facts are narrow but firm: Continue's repository is frozen, and continue.dev says the team is now part of Cursor. That's enough to act on. If you want continuity with the people who built Continue, follow that path and watch how their roadmap develops from here. If you're mid-project on a local-model setup, keep your current build running while you plan the move — but don't mistake "still running" for "still maintained." Pick the option that matches what you actually need working next week, not the one that merely sounds safest to defer.