Launching today
Codemind
Delegate coding. Get back code that works.
41 followers
Delegate coding. Get back code that works.
41 followers
Your coding agent hands Codemind a task over MCP. It writes the code, tests it against your criteria and returns the files. $0.10 per verified build, failed builds free, free allowance to start.
Launch Team
Viktor.comAn AI coworker that actually does the work
Promoted
Hi Product Hunt, I'm Andriy, and I built Codemind.
If you code with an agent like Claude Code, Cursor, Codex or OpenCode, you know the pattern: the agent spends your plan's tokens on routine tasks you could hand off. Codemind is where it hands them off. The agent sends a task and acceptance criteria over MCP, Codemind writes the code, tests it against those criteria, and returns the files. You are billed only for a build that comes back verified. A failed build is free.
What you can do today:
- Delegate a single task from your agent and get files back with the test result.
- Let your agent set itself up: paste one prompt and it creates a free account and connects.
Coming next: Swarm (many tasks at once, Pro).
I'd love to hear which of your own tasks you'd hand off first, and which you wouldn't. I'm in the thread all day.
Can I set my own testing criteria so the returned code matches my project's requirements?
Yes, Varun. You write the acceptance criteria yourself, and they travel with the task. Codemind tests the code against them and returns the test result with the files. A build that doesn't come back verified is free. What kind of requirements do you have in mind for your project?
Just to make sure I understand correctly, the benefit here is that once you have a plan designed by the coding agent, codemind can execute the individual plan steps in parallel? Speeding up build time and saving tokens?
Is this at the function level? module level? or any scope of task?
@simon_moxon Hey Simon, thanks for your question!
Speeding up build time and saving tokens - and also offloading all the work to the cloud, so your laptop doesn't crash when you code 100s projects in parallel as I do :)
Regardless if you have a plan or not, your main AI coding agent such as Claude Code/Codex/Cursor/OpenCode/etc will select coding tasks suitable to delegate to Codemind, spec them and then check the work product before merging.
Any scope of task really, from pure function to a full product spec - you agent will do it automatically, or you can tune your preferences to pick and choose specific type of coding work.
It can do coding, code reviews, testing, etc etc etc
You can send as many tasks as you want in parallel, restricted only by your plan and usage credit.
Happy to explain more!
@simon_moxon Perhaps easier parallel - think of Claude Code/Codex and Codemind as a Engineering Manager+TeamLead and a Dev relationship :)
Andriy, I like that the acceptance criteria go in with the task and the test results come back with the files. That gives the calling agent something concrete to check. Congrats on the launch!
Thanks, Oleksii. That's the idea: the agent can check the result against its own criteria instead of taking my word for it. Which kind of task would you try handing off first?