Comment by silcoon
Comment by silcoon 20 hours ago
Why did you remove AI agent configurations and instructions from the repo? See .gitignore
Comment by silcoon 20 hours ago
Why did you remove AI agent configurations and instructions from the repo? See .gitignore
Fair point - I should be more transparent. Yes, Claude assisted significantly with development. The .gitignore excludes AI config files because they where not needed in the project and aren't useful to others. I'll add a note to the README about AI-assisted development. The code is reviewed and understood, not blindly accepted.
Thanks for your reply. Mine wasn't a critique but a genuine curiosity. I was interested to see what where the base instructions used for a rust project.
> The .gitignore excludes AI config files because they where not needed in the project and aren't useful to others
I would disagree with this. Since it's an open-source project it would be beneficial to everyone, especially to future contributors, to agree in good code practices and conventions when using LLMs. I would say they're really useful.
You raise a fair point, and I've thought about this a bit since you mentioned it. What those files actually are:
The files you noticed (AGENTS.md, GEMINI.md, CLAUDE.md, etc.) are auto-generated scaffolding from Task Master AI, a task management tool I use. They contain MCP (Model Context Protocol) configuration for various AI coding assistants — essentially boilerplate instructions for how different AI tools should connect to the task system. They're not really "coding conventions" so much as tool-specific wiring.
That said, you're right that it would be useful to share the actual conventions and rules I use, especially for an AI-generated project.
What I'd be happy to share:
My Cursor system prompt — I have a general coding principles prompt I use across projects. It's not Ferrite-specific but covers code style, architecture preferences, etc. Happy to share this.
Project-specific rules — I do have .cursor/rules/ files with Task Master workflow conventions. These could be included if contributors want to use similar AI-assisted workflows.
The actual workflow — How I structure tasks, iterate on implementations, and the back-and-forth with AI. This might be more useful than config files.
Why I excluded them initially: Honestly, most of those files are boilerplate MCP configs that would only matter if someone wanted to use the exact same tooling stack (Cursor + Task Master + specific MCP servers). I wasn't sure that would be useful to most contributors, and I didn't want to clutter the repo with tool-specific noise. But I take your point about setting conventions for AI-assisted contributions.
Let me know if there's interest, and I can:
Add a CONTRIBUTING_AI.md or similar with the actual conventions and workflow
Share my system prompt in the docs
Un-ignore some of the rules files if they'd help
What would be most useful to you?
Could you estimate how much was written by AI vs you? Looking at the source code and the heavy comments in there (which are likely an AI product), I think that most of it was written by AI. Same with the whole docs directory.
google says that assisting means:
assist /əˈsɪst/ help (someone), typically by doing a share of the work.
So in this case... wouldn't the relationship be inverted, e.g. you assisting AI? (semi joking ;))
You're right to push on this, let me be fully transparent.
100% of the code was generated by AI (Claude Opus 4.5(I am super impressed by the capabilities of Opus 4.5), via Cursor with MCP tools). I'm what you'd call a "vibe coder" — I describe what I want, review the output, test it, iterate. I haven't written Rust by hand for this project.
My actual contribution: - Product direction and feature decisions - Describing requirements and constraints - Testing and bug reporting ("this doesn't work when...") - Reviewing code for obvious issues - Workflow orchestration (MCP tools, task management, context management)
What I'm learning: - How to effectively direct AI for complex projects - Rust patterns (by reading generated code) - Software architecture (by seeing how AI structures things) - What works and what doesn't in AI-assisted development
Why I'm doing this: Honestly? To learn and experiment. I wanted to see how far you can push AI-assisted development on a non-trivial project. Ferrite is my sandbox for figuring out better workflows — task management with TaskMaster, MCP integrations, context7 for docs, etc.
Is this "real" software development? I don't know. It's definitely a new paradigm. The code compiles, runs, and does useful things. Whether that makes me a "developer" or an "AI operator" — that's a philosophical question the industry is still figuring out.
The documentation and comments being AI-heavy was a fair tell. I probably should have been upfront about this from the start.
Please respect the HN community and kindly disclose when you are using an LLM to respond to user feedback.
It's vibe coded. The entire project is only 10 commits, a few of them are giant with a bunch of markdown files full of emojis in the docs/ folder.