Project ideas from Hacker News discussions.

I built a programming language using Claude Code

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

4Dominant Themes in the Discussion

Theme	Key Takeaway	Illustrative Quote
1. LLMs can write whole codebases unsupervised	A developer let Claude generate every line of a project and only added guardrails afterwards.	“While working on Cutlet, though, I allowed Claude to generate every single line of code. I didn’t even read any of the code. Instead, I built guardrails to make sure it worked correctly…” – andsoitis
2. Language choice still matters for correctness & performance	Even when a model writes code, a language’s guarantees (speed, safety, error handling) affect the final product.	“A language which offers certain correctness guarantees may still be the most efficient way to build a particular piece of software even when it's a machine writing the code.” – marssaxman
3. LLMs can adopt brand‑new or obscure languages with minimal exposure	A DSL with no public docs was understood and used by Codex/Claude just from the repository’s context.	“There is certainly nothing in any AI's training data about it. And yet: codex has no trouble reading my repo, understanding how my DSL works, and generating code written in this novel language.” – marssaxman
4. Token‑efficiency is driving interest in terse, AI‑friendly languages	Researchers speculate that future languages may be designed more for LLM parsing than for human readability.	“I think token efficiency will continue to be a problem. So you could imagine very terse programming languages that are roughly readable for a human, but optimized to be read by LLMs.” – danielvaughn

🚀 Project Ideas

[AI Guardrails DSL]

Summary

A lightweight DSL that lets developers declaratively specify correctness constraints, invariants, and automatically generate comprehensive test suites for AI‑produced code.
Provides safety layers so LLM‑generated code can be trusted without manual review.

Details

Key	Value
Target Audience	AI‑centric developers, security‑focused teams
Core Feature	Declarative safety contracts with auto‑generated unit and property tests
Tech Stack	Rust compiler, WASM sandbox, Python test harness, GitHub Actions CI
Difficulty	Medium
Monetization	Hobby

Notes

HN users repeatedly stress the need for guardrails and automatic testing when relying on AI code (e.g., “guardrails to make sure it worked correctly”).
Solves the practical pain of debugging AI output by embedding verification directly into the language.

[ContextPilot]

Summary

A CLI that automatically builds and maintains an optimized context file for LLMs by relevance‑scoring repository files and compressing them into a token‑efficient snapshot.
Eliminates manual context management, letting users focus on prompting rather than curating inputs.

Details

Key	Value
Target Audience	LLM‑assisted developers, startup engineers
Core Feature	Dynamic context builder with relevance scoring and up‑to‑date .context file generation
Tech Stack	Node.js, TypeScript, SQLite, OpenAI API
Difficulty	Low
Monetization	Revenue-ready: SaaS subscription

Notes

Commenters lament the “context window” headache and token cost when feeding large codebases to LLMs.
Directly addresses the practical utility of keeping relevant snippets in context for sustained productivity.

[TokenFit]

Summary

A minimal, token‑efficient programming language designed specifically for LLM ingestion, featuring ultra‑short syntax and built‑in test scaffolding.
Enables LLMs to generate correct code faster while remaining readable enough for human sanity checks.

Details

Key	Value
Target Audience	AI‑first researchers, hobby language creators
Core Feature	Token‑optimized grammar with optional type hints and auto‑generated stub tests
Tech Stack	Ruby parser generator, Rust compiler backend, Markdown test templates
Difficulty	High
Monetization	Hobby

Notes

HN discussions highlight that terse languages may actually improve LLM performance and that “token efficiency” is a recurring concern.
Offers a concrete solution by providing a language whose design goals align with LLM strengths.

[LLM Language Marketplace]

Summary

A SaaS platform where creators upload DSL specifications and the system instantly generates AI‑produced libraries, complete test suites, and sandboxed executors.
Turns niche languages into immediately usable, AI‑maintained packages with versioning and discoverability.

Details

Key	Value
Target Audience	Language designers, tooling startups
Core Feature	Spec upload → AI codegen + automated testing + Docker sandbox deployment
Tech Stack	Django + FastAPI, PostgreSQL, Docker, GitHub Actions
Difficulty	Medium
Monetization	Revenue-ready: Pay‑per‑generation credits

Notes

Frequent HN interest in “personal unique languages” and the difficulty of getting LLMs to adopt them without massive training data.
Provides a practical utility by handling distribution, testing, and execution, turning a theoretical pain point into a marketable service.