https://github.com/russellgn/ai-codelint
Use LLMs to lint code.
https://github.com/russellgn/ai-codelint
ai async-openai code-quality-analyzer generative-ai ide-extension language-server-protocol linter lsp openrouter rust-lang tower-lsp typescript vscode
Last synced: 23 days ago
JSON representation
Use LLMs to lint code.
- Host: GitHub
- URL: https://github.com/russellgn/ai-codelint
- Owner: RussellGN
- License: mit
- Created: 2026-03-24T07:00:09.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2026-05-03T16:39:18.000Z (24 days ago)
- Last Synced: 2026-05-03T18:31:37.520Z (24 days ago)
- Topics: ai, async-openai, code-quality-analyzer, generative-ai, ide-extension, language-server-protocol, linter, lsp, openrouter, rust-lang, tower-lsp, typescript, vscode
- Language: Rust
- Homepage: https://russellgn.github.io/AI-CodeLint/
- Size: 1.77 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- License: LICENSE
Awesome Lists containing this project
README
# AI CodeLint
Use LLMs to lint code.
AI CodeLint uses LLMs to identify logic bugs that survive compilation and traditional linting.
## Install (Recommended)
macOS/Linux:
```sh
curl -fsSL https://raw.githubusercontent.com/RussellGN/AI-CodeLint/main/scripts/install.sh | sh
```
Windows (PowerShell):
```powershell
irm https://raw.githubusercontent.com/RussellGN/AI-CodeLint/main/scripts/install.ps1 | iex
```
After install:
```sh
ai-codelint --version
```
## Development Status
AI CodeLint is in heavy development and is not production-ready yet.
- Features and APIs can change quickly.
- Breaking changes are expected while core behavior is refined.
- Current releases should be treated as pre-release testing builds.
## Interface
AI CodeLint consists of one standalone binary: `ai-codelint`.
That binary currently exposes two modes of operation via the command line:
- CLI mode: run lint checks from the terminal, like conventional linters.
- LSP server mode: run as a language server so IDE clients can request diagnostics.
The VS Code extension is now live on the Visual Studio Marketplace:
- [AI CodeLint for VS Code](https://marketplace.visualstudio.com/items?itemName=russell-gn.ai-codelint)
Other IDE extensions are incoming.
## Inference
AI CodeLint intentionally uses [OpenRouter](https://openrouter.ai/) so inference-provider and model selection is quick and easy to change and experiment with.
- **You will need an [OpenRouter api key](https://openrouter.ai/keys) to use AI CodeLint**.
- At the time of writing, signing up is free, with free models available for use with limits.
- Run `ai-codelint --configure` to set up api key and other optional runtime defaults.
Model quality note:
- Best results so far come from frontier models.
- Lesser models can produce highly variable lint quality and consistency.
- Recommended model: `anthropic/claude-sonnet-4.6`
## Examples
This repository includes intentionally buggy code samples under [`lint samples/`](https://github.com/RussellGN/AI-CodeLint/tree/main/lint%20samples/).
These samples cover:
- syntax/semantics mistakes (should be ignored by linter),
- easy rule-based linter catches,
- subtle logic bugs,
- edge-case behavior and review-evasive patterns.
They are useful for evaluation, and comparing model performance.
Resources:
- Official Docs (still in development): [russellgn.github.io/AI-CodeLint](https://russellgn.github.io/AI-CodeLint)