Gnista keeps an eye on Claude Code, Codex, Gemini CLI and every other coding agent you start, in your MacBook’s notch or menu bar, and tells you the moment one needs you.
I’m Daniel, a designer with twenty years in product design and design systems. I’m not a Mac developer, and Gnista is the first native app I’ve shipped.
The problem: I run Claude Code and Codex side by side across client projects, some on my laptop and some on servers over SSH. An agent would wait ten minutes for one approval I never saw, while I was deep in a Figma file somewhere else.
What Gnista does: it sits in your MacBook’s notch (or in the menu bar if you prefer) and keeps track of every running agent. When one asks, the notch opens with the whole request. Allow or deny it right there, or click to land in the exact terminal tab, tmux pane or editor window.
A few things I care about:
Buy it once and keep it, with an optional yearly plan to keep updates coming. Not in the App Store, no account needed.
28 coding agents, plus any tool with hooks. I have about 30 CLIs installed just to test against.
Works over SSH, where a missed prompt costs the most.
Six languages from day one, including Chinese, Japanese and Korean.
My first idea was a notch that glowed while agents worked. I dropped it. A quiet notch that only speaks up when something is waiting turned out far more useful.
What would make you trust an agent to run while you look away? I’d love to hear it.
Surfacing the moment an agent needs me in the notch is the right place for it, because I keep three agents running and mostly end up watching a spinner instead of doing the work I meant to do.
the less flashy version is the right call, anything that blinks at me all day gets ignored by tuesday
About Gnista on Product Hunt
“Run more coding agents, watch none of them”
Gnista launched on Product Hunt on November 12th, 2026 and earned 73 upvotes and 3 comments, placing #22 on the daily leaderboard. Gnista keeps an eye on Claude Code, Codex, Gemini CLI and every other coding agent you start, in your MacBook’s notch or menu bar, and tells you the moment one needs you.
Gnista was featured in Artificial Intelligence (483.5k followers), Menu Bar Apps (12.3k followers) and Apple (15.6k followers) on Product Hunt. Together, these topics include over 147.9k products, making this a competitive space to launch in.
Who hunted Gnista?
Gnista was hunted by Daniel Zander. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how Gnista stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
Daniel Zander
Hi Product Hunt 👋
I’m Daniel, a designer with twenty years in product design and design systems. I’m not a Mac developer, and Gnista is the first native app I’ve shipped.
The problem: I run Claude Code and Codex side by side across client projects, some on my laptop and some on servers over SSH. An agent would wait ten minutes for one approval I never saw, while I was deep in a Figma file somewhere else.
What Gnista does: it sits in your MacBook’s notch (or in the menu bar if you prefer) and keeps track of every running agent. When one asks, the notch opens with the whole request. Allow or deny it right there, or click to land in the exact terminal tab, tmux pane or editor window.
A few things I care about:
Buy it once and keep it, with an optional yearly plan to keep updates coming. Not in the App Store, no account needed.
28 coding agents, plus any tool with hooks. I have about 30 CLIs installed just to test against.
Works over SSH, where a missed prompt costs the most.
Six languages from day one, including Chinese, Japanese and Korean.
My first idea was a notch that glowed while agents worked. I dropped it. A quiet notch that only speaks up when something is waiting turned out far more useful.
What would make you trust an agent to run while you look away? I’d love to hear it.