Show HN: Deslop – Oxlint rules for AI-generated TypeScript

📝 Discussion Summary (Click to expand)

Three prevalent themes

Rule quality boosts LLM output – Users stress that solid rule sets noticeably improve LLM results.

aleksiy123: “I really believe having good rules like this can make a big difference in llm quality.”
Adoption of specific linting/formatting tooling – Practitioners combine existing linters (Oxlint, Vale) with custom AST‑based tools to enforce cross‑language rules.

aleksiy123: “I’m personally using both oxlint for typescript and my own vibe coded tool for doing cross language rules ast rules and llm hook outputs. And recently added vale as well.”
Community contribution and sharing – Encouragement to submit PRs and spread useful rules to others’ codebases.

maharshi365: “Awesome. Feel free to make a pr for any rules that you think are important. I've started adding to my friends sloppy codebases and its great!”

🚀 Project Ideas

Generating project ideas…

A pluggable framework that lets developers define linting rules once in a simple DSL and automatically generate adapters for tools like oxlint, Vale, eslint, and custom LLM hook outputs, eliminating the need to write separate rule implementations for each language or linter.
Core value: write once, enforce everywhere, reducing duplication and keeping LLM-generated code and prompts consistent across stacks.

Key	Value
Target Audience	Teams building LLM-powered developer tools, DevOps engineers enforcing AI output quality
Core Feature	DSL → multi‑linter adapters (oxlint, Vale, eslint, custom AST hooks)
Tech Stack	Rust core, WASM plugins, Node.js bindings, Tree‑sitter grammars
Difficulty	Medium
Monetization	Revenue-ready: SaaS tier for hosted rule management + CI integration ($20/mo per seat)

HN users praised “my own vibe coded tool for doing cross language rules ast rules and llm hook outputs” (aleksiy123) – this product turns that vibe‑coded effort into a reusable engine.
Enables easy PR contributions of new rules (as maharshi365 invited) and fuels discussion on rule sharing and LLM quality improvements.

A hosted API and CLI that runs configurable linting rule sets (style, safety, hallucination detection) on LLM‑generated text or code and returns annotated feedback, ready to plug into CI pipelines or editor extensions.
Core value: instant, consistent quality checks for AI output without teams having to maintain their own linting infrastructure.

Key	Value
Target Audience	AI product teams, developers using LLMs for code generation, content creators leveraging LLMs for writing
Core Feature	API‑driven linting with pre‑built rule packs (Vale, oxlint, custom) and webhook/GitHub Action integration
Tech Stack	Python FastAPI, Docker, Redis for rule caching, Lark parser for DSL, React dashboard
Difficulty	Medium
Monetization	Revenue-ready: Usage‑based pricing ($0.001 per lint request, free tier 10k/month)

aleksiy123’s comment about using “oxlint for typescript” and “vale” shows a desire to combine multiple linters for LLM output – this service unifies them under one API.
Provides a concrete platform for discussing rule effectiveness and encourages community rule contributions, mirroring the PR invitation in the thread.

A public registry (similar to npm) where developers can publish, version, and install LLM‑focused rule packs (e.g., “llm-security”, “llm-style-guide”) and consume them via a CLI that integrates with oxlint, Vale, or custom hook systems.
Core value: simplifies sharing and reuse of rule sets, turning ad‑hoc vibe‑coded tools into collaboratively maintained assets.

Key	Value
Target Audience	Open‑source maintainers, enterprise AI teams, tool builders wanting to distribute LLM linting rules
Core Feature	Registry + CLI to install/update rule packs, with semantic versioning and dependency resolution
Tech Stack	Node.js/Express backend, PostgreSQL for package metadata, GraphQL API, CLI in TypeScript
Difficulty	Low
Monetization	Hobby

maharshi365’s invitation to “make a pr for any rules that you think are important” directly aligns with a hub that welcomes PR‑style contributions to rule packages.
Sparks discussion on rule quality, versioning, and cross‑language applicability, potentially becoming a central community resource like DefinitelyTyped for LLM linting.