Gemini CLI has crossed the line from experiment to fast-moving developer infrastructure. ToolVitals records 103,627 GitHub stars, 30 releases in 90 days, and 18 release events in the last 30 days. That is not quiet maintenance. That is a product team pushing hard on a terminal-native AI agent.
The official site positions Gemini CLI around building, debugging, deploying, querying and editing large codebases, generating apps from images or PDFs, and automating workflows from the terminal with Gemini 3. The GitHub README uses the same frame: an open-source AI agent that brings Gemini into the command line.
The notable signal: release cadence with agent plumbing
The interesting part is not just the star count. Stars can spike from launch attention. The release stream is harder to fake.
The April 28 nightly release included changes around custom theme schema validation, manual update guidance when automatic updates fail, better ECONNRESET and ETIMEDOUT error messages, MCP list command behavior, sandbox proxy cleanup, session UUID handling, and new ContextManager and AgentChatHistory wiring. That is plumbing work. It points to a tool being hardened for daily use, not just polished for screenshots.
The payload also shows nearby releases touching workspace trust in headless mode, secure .env loading, loop detection crashes, OpenSSL streaming retries, YOLO mode behavior, bundled ripgrep binaries for offline support, memory commands, and performance test coverage. Taken together, the bet is clear: Gemini CLI wants to be useful in real developer environments, including scripted, headless, offline-ish, and failure-prone ones.
That matters because terminal AI agents live or die on boring details. Authentication errors, sandbox cleanup, network timeouts, MCP timeouts, and update failures are not glamorous. They are exactly the stuff that decides whether engineers keep the tool installed after the first week.
Popular, but still competing on execution
ToolVitals gives Gemini CLI a 229.2 hot score, 95 health score, 100 shipping score, and 94 overall ToolVitals score. The GitHub page currently rounds the repository display to 104k stars, while the ToolVitals snapshot records 103,627. I am keeping the ToolVitals number because the payload is the source of truth for metrics.
The related-tools comparison puts Gemini CLI in a heavy group. LangChain has 136,337 stars, a 240.0 hot score, and 32 release events in 30 days. OpenClaw shows 370,596 stars, a 239.2 hot score, and 44 release events in 30 days. Gemini CLI is not the loudest project by raw activity, but it is already operating at top-tier shipping velocity for developer tools.
ToolJet is a closer release-rate comparison: 37,882 stars, 229.1 hot score, and 15 release events in 30 days. Gemini CLI has nearly the same hot score, more stars, and a slightly higher 30-day release count. That says the market is paying attention, and the maintainers are still shipping.
What ToolVitals cannot infer
ToolVitals can see public signals: stars, release events, score movement, website availability, SSL status, and source-linked activity. It can also connect those signals to first-party positioning and release notes.
ToolVitals cannot prove code quality, user satisfaction, retention, revenue, model quality, or whether Gemini CLI works better than a rival agent on your codebase. It cannot tell whether a release fixed the bug that annoyed your team last week. It can only say the project is alive, heavily watched, and moving quickly.
That distinction matters here. The public evidence supports the claim that Gemini CLI is being actively developed and positioned as a serious terminal AI agent. It does not prove that it should replace your current coding agent.
Recommendation
If your team already works in terminals, scripts, and GitHub workflows, evaluate Gemini CLI now, specifically against your real repo automation tasks. Do not test it with toy prompts. Test update behavior, MCP integration, sandboxing, headless runs, large-codebase edits, and failure recovery.
The data says Gemini CLI is moving fast enough to deserve that trial. The release notes say the team is fixing the unsexy problems that make terminal agents stick.