Launching today

SpecBuddy

Launching today

Control your AI coding agent — from plan to every step

8 followers

Control your AI coding agent — from plan to every step

8 followers

Visit website

Chat-coding lets you review code after it's written. SpecBuddy lets you review the plan before: a spec, a plan, then step-by-step execution you can accept, reject, or roll back — in your JetBrains IDE, on top of Claude Code or Codex. Free in public beta.

Free

Launch tags:Productivity•Developer Tools•Vibe coding

Launch Team / Built With

Viktor.comAn AI coworker that actually does the work

Promoted

Maker

📌

Hi Product Hunt 👋 I'm Alex. I've spent ~10 years building developer tools on the JetBrains platform, and the last year watching my team — and myself — get frustrated with AI coding agents in exactly the same way. The pattern is always the same. You write a prompt. The agent generates 200 lines. You read the diff and realize it misunderstood you three turns ago. So you prompt again. And again. The problem isn't that we're bad at prompting. It's that we only get to review the work AFTER it's done. Code review is the wrong place to catch "you built the wrong thing." SpecBuddy moves the review earlier. You describe what you want, the agent turns it into a spec and a plan, and you approve that plan before a single line of code is generated. Then it executes step by step — and every step is something you can accept, reject, or roll back. If you don't want to write specs at all, that's fine too: run your agent in plain chat and SpecBuddy still catches every change as a reviewable diff you can comment on and send straight back to the agent. Honest scope on day one: — It installs across the JetBrains family (IDEA, PyCharm, GoLand, WebStorm, Rider, CLion and the rest), 2026.1+. But we built and tested it in IntelliJ IDEA — the others should work, and I'd love to know if they don't. — Works on top of Claude Code and Codex. We're not another model. — VS Code and Cursor are next. — Free during public beta. No billing, no card. I'd genuinely like to hear where this breaks for you. I'm here all day.

Report

1d ago

Finally a tool that addresses the actual pain point with chat-coding. Tried it on a refactor in IntelliJ and the step-by-step accept/reject flow made it feel way safer than just trusting whatever Claude spits out.

Report

5h ago

Tried it with a refactor on a messy Kotlin module and being able to reject individual steps before they run feels way less scary than watching an agent go wild. The rollback is what sold me.

Report

5h ago

Love that you made the spec the checkpoint instead of the code itself. Honestly feels way more natural to review intent before execution, and the rollback per step is a really thoughtful touch.

Report

3h ago

Viktor.comAn AI coworker that actually does the work

Promoted

Maker

📌

Report

1d ago

Report

5h ago

Tried it with a refactor on a messy Kotlin module and being able to reject individual steps before they run feels way less scary than watching an agent go wild. The rollback is what sold me.

Report

5h ago

Love that you made the spec the checkpoint instead of the code itself. Honestly feels way more natural to review intent before execution, and the rollback per step is a really thoughtful touch.

Report

3h ago