Hey Product Hunt,
I’m Renald, the founder of Cohesium AI.
I created Assemble because I was tired of AI tools that seem helpful but remain generic. Code review becomes a polite summary. A security audit becomes a reformatted checklist. A multistep project starts strong, then falls apart as the context lengthens or the task becomes more complex.
So I created what I actually needed: A structured AI work systemnot just another assistant.
With Assemble, you type /go and specify what you need. From there, it sets the task hard, staying useful. Cross-session memoryand switches to a. Feature-based workflow When the task is complex. For larger deliverables, it can also take implementation across the board with review and test phases.
What makes it different from most agent frameworks:
• This is a configuration generator, not a runtime.
• Zero daemon, zero SDK, zero dependencies, zero lock-in
• Local configurations for 21 platforms including Cursor, CloudCode, Codex, Gemini CLI, Copilot, and Windsurf.
• It also works beyond coding: Documents, contracts, proposals, email, and client operations
Marvel Framework isn’t branding – it’s a quick engineering choice. In testing, this gave us stronger character recognition, better consistency, and less common output than traditional agent setups.
And because LLMs naturally agree so easily, assemble backs into structured disagreements: Deadpool challenges assumptions by default, and Dr. Doom escalates high-stakes decisions.
A real turning point for me: a client project that was supposed to take 2 days turned into 10 days of failed attempts with typical AI tools. With Assemble, it took 30 minutes.
If you try it, I’d really love your feedback — especially on the workflows, platforms, and expert roles you want next.
MIT Licensed. Open source. Built for real work.