|
|
|
|
Show HN: Requirements Engineering with Formal Verification
(fizzbee.ai)
|
|
27 points
by jayaprabhakar
21 days ago
|
|
|
I've been building an open source formal methods system (fizzbee.io) for the past few years. Today I'm launching a new app built on the same technology. It performs requirements engineering using formal verification to uncover gaps and produce precise instructions for your coding agents to follow. When given a prompt, it
- asks high signal follow-up questions
- converts to formal spec and identifies complex requirements gaps
- generates validation scenario At the end, it produces a specification document that can be shared with coding agents. In my trials on various projects, it produces working code in fewer iterations. Please give it a try and share your feedback.
https://fizzbee.ai/ You can also look at a sample project.
https://fizzbee.ai/projects/94bf2869-97a1-445c-8f5d-4445848b... |
|
|
I would add a few points to consider:
1. I think it would be better as a VS Code / Cursor extension, or some other local tool. That's because I would want it to know my domain: see existing markdown files, specs, review existing code, available tools, concepts, terminology, and more. While prompting that fizzbee.ai tool, I want my domain to be discoverable, searchable, and link the important items manually. Like the usual work with AI agents.
2. I think it should come with skills or mcp servers, to instruct AI agents to execute the spec. Having a markdown file is nice, but I think the formal part should be in fizzbee files, referenced by the informal spec, so the AI agents will be able to iterate on it, execute, compare, drive, and if needed, even update it (after discovering inconsistencies in the code).
So, it looks like a very promising MVP and the right direction, but there so much more (existing) stuff to add!