Nothing built yet, but I was wondering if you’ve also run into the problem of keeping strong, opinionated guidelines when working with LLMs.
I’m thinking about an open-source solution to help teams define and share high-level project guidelines (PEP8, architectural rules, naming conventions, product principles, tone of voice, etc.) in a way that’s actually usable by LLMs.
The idea would be simple:
- a few clicks to configure your guidelines.
- you get a public URL with LLM-readable content and a YAML file you can download and plug directly into Cursor/Copilot/Google Cli/other tools
The goal is that teams can have shared, explicit, opinionated guidelines that stay consistent across prompts, editors, and contributors. without copy-pasting the same instructions everywhere or relying on tribal knowledge.
Think:
- onboarding new devs faster.
- less “LLM drift” over time.
- clearer defaults for AI-assisted code and writing
- guidelines that live next to the project and evolve with it.
So, very early idea, but curious:
- does this resonate with problems you’re seeing?
- what kind of rules or constraints would you actually want to encode?
- anything like this you’ve tried before (and why it didn’t stick)?
Nothing built yet, but I was wondering if you’ve also run into the problem of keeping strong, opinionated guidelines when working with LLMs.
I’m thinking about an open-source solution to help teams define and share high-level project guidelines (PEP8, architectural rules, naming conventions, product principles, tone of voice, etc.) in a way that’s actually usable by LLMs.
The idea would be simple:
- a few clicks to configure your guidelines.
- you get a public URL with LLM-readable content and a YAML file you can download and plug directly into Cursor/Copilot/Google Cli/other tools
The goal is that teams can have shared, explicit, opinionated guidelines that stay consistent across prompts, editors, and contributors. without copy-pasting the same instructions everywhere or relying on tribal knowledge.
Think:
- onboarding new devs faster.
- less “LLM drift” over time.
- clearer defaults for AI-assisted code and writing
- guidelines that live next to the project and evolve with it.
So, very early idea, but curious: - does this resonate with problems you’re seeing? - what kind of rules or constraints would you actually want to encode? - anything like this you’ve tried before (and why it didn’t stick)?
Happy to hear thoughts or pushback.