AI CodeLint VS Code Extension

Boost code quality as you type.

This extension runs the ai-codelint language server in the background and surfaces AI-assisted logic diagnostics directly in VS Code.

Project Status

AI CodeLint is in heavy development and not production-ready yet.

Features and behavior can change quickly.
Breaking changes are expected while core behavior is refined.

What This Extension Does

Starts ai-codelint --mode server as an LSP server.
Analyzes open files and untitled buffers (file and untitled URI schemes).
Reports likely logic bugs that can survive compilation and traditional linting.

Prerequisites

Install the ai-codelint binary.
Set OPENROUTER_API_KEY as a global environment variable.
Keep your extension/binary version aligned with the recommended version in status.json.

Install The Binary (Recommended)

macOS/Linux:

curl -fsSL https://raw.githubusercontent.com/RussellGN/AI-CodeLint/main/scripts/install.sh | sh

Windows (PowerShell):

irm https://raw.githubusercontent.com/RussellGN/AI-CodeLint/main/scripts/install.ps1 | iex

Then verify installation:

ai-codelint --version

Configure Runtime Defaults

Run the CLI walkthrough once to set defaults such as model and max output tokens, and to get API key setup guidance:

ai-codelint --configure

The root README currently recommends anthropic/claude-sonnet-4.6 for best lint quality.

Install The VS Code Extension

Install from the Visual Studio Marketplace:

AI CodeLint (russell-gn.ai-codelint)

Or install via CLI:

code --install-extension russell-gn.ai-codelint

Option A: Run In Extension Development Host (Local Development)

From the repository root:

cd server && cargo build
cd ../clients/vs-code && pnpm i && pnpm compile

Then in VS Code, run the launch config named Launch VS Code Client (F5). The workspace launch configuration already sets:

SERVER_PATH=${workspaceRoot}/server/target/debug/ai-codelint

Option B: Package As VSIX And Install Locally (Unreleased Builds)

From clients/vs-code:

pnpm i
pnpm compile
pnpm package

Install the generated .vsix in VS Code using:

Extensions: Install from VSIX...

Settings

The extension currently contributes one setting:

ai-codelint.trace.server
- off: no trace logging
- messages: error-only trace output (default)
- verbose: full trace output

Use the AI CodeLint trace output channel in VS Code when troubleshooting.

Environment Variables

OPENROUTER_API_KEY (required): API key used for lint inference.
SERVER_PATH (optional): path to a custom ai-codelint binary. If not set, the extension uses ai-codelint from your PATH.

Important: after changing environment variables globally, fully restart VS Code so the extension host picks them up.

Troubleshooting

"OPENROUTER_API_KEY environment variable is required"

Set OPENROUTER_API_KEY globally for your OS/shell.
Restart VS Code completely.

"Current version 'x' of ai-codelint is out of date"

Update your local setup to the latest recommended release.
Ensure extension and binary versions are up to date.

Server Does Not Start

Verify ai-codelint --version works in a terminal.
If using a custom path, verify SERVER_PATH points to an executable binary.
Set ai-codelint.trace.server to verbose and check the AI CodeLint trace channel.

Development

From repository root:

cd server && cargo build && cd ../clients/vs-code && pnpm i && pnpm compile

There is also a workspace task named install and compile that runs the same workflow.

Useful Scripts (`clients/vs-code`)

pnpm compile: type-check and bundle extension output
pnpm watch: watch mode for rapid extension iteration
pnpm lint: run oxlint
pnpm fmt: run oxfmt --write
pnpm package: build a .vsix artifact

ai-codelint

Russell GN