Continue’s public face has changed. The website now opens with “Quality control for your software factory,” and the repo README says “Source-controlled AI checks, enforceable in CI.” That is a loud pivot for a project with 32.8k stars, 21,498 commits, 843 tags, 30 releases in 90 days, and 13 release events in 30 days.
The real bet
This is not being framed as a generic coding copilot. It is being framed as a gate in the delivery pipeline. Checks live in .continue/checks/, run as GitHub status checks, and can suggest diffs when code misses the mark. The docs and README point people to a walkthrough, a CLI install, and then cn. That says the team wants the center of gravity in repo-local policy, not chat-window convenience.
The release notes back that up. The Mar. 27 VS Code prerelease added .continue/configs support, fixed config.yaml creation and population, and tightened session-history filtering by workspace directory. The JetBrains prerelease on the same day added the same config fix, plus a logger redirect to stderr to avoid IPC stream corruption. Those are not vanity changes. They are the kind of fixes you make when the tool has to behave predictably inside editors and CI.
What the data does not tell you
ToolVitals can see stars, releases, and commit activity. It cannot see code quality, model quality, user satisfaction, revenue, or whether teams actually trust the checks in production. It also cannot prove that the CI-first pitch is better than the older editor-assistant story. It only shows that the project is actively shipping and tightening the plumbing around a specific workflow.
Comparison with nearby tools
Continue sits in the same busy class as ToolJet, Composio, LangChain, OpenClaw, and Gemini CLI. At 32.8k stars, it is close to ToolJet’s 37.8k and above Composio’s 27.9k, but far below LangChain’s 134.6k and OpenClaw’s 362.7k. On release cadence, Continue’s 13 release events in 30 days trails Gemini CLI’s 33 and sits below ToolJet’s 27 and LangChain’s 27. The takeaway is simple, it is active, but not manic.
Bottom line
If your team wants AI in review, not just in autocomplete, Continue deserves a trial. If you only want an editor sidekick, the current messaging says this project has moved past that center of gravity.