Gatekeep

Gatekeep is a local web app for reviewing codebases with an OpenAI-compatible model.
It runs on your machine, keeps project/chat history locally, and helps you initialize and inspect repositories safely.

What You Get

• Local browser UI for project setup and chat-based review.
• One-click project initialization that creates docs/intro.md in the selected repo.
• Project markdown context loading, including special handling for developer.md.
• Built-in Doctor checks for Node runtime, API config, and project path readiness.
• Local storage for project history, tool traces, token usage, and estimated cost.

Requirements

• Node.js 20+
• An OpenAI-compatible API key

Install

Install from npm (recommended):

npm install --global @aakrit512/gatekeep

Or run from source:

pnpm install
pnpm run build
pnpm link --global

Start Gatekeep

gatekeep start --project /path/to/repo

Defaults:

• Host: 127.0.0.1
• Port: 9808

You can also run:

gatekeep start --project .
gatekeep start --port 9809 --host 0.0.0.0

Then open http://127.0.0.1:9808.

First-Run Setup (In the UI)

1. Choose a project directory.
2. Open Config and set:
   • AI_API_KEY
   • AI_BASE_URL (for example https://api.openai.com/v1)
   • MODEL_NAME (for example gpt-4o)
3. Run Doctor and fix any reported issues.
4. Click Initialize to generate docs/intro.md.
5. Start chatting in the review panel.

Project Markdown Context

Add a developer.md file at the root of any project you review. Put developer-specific notes there: coding standards, architecture constraints, review preferences, deployment caveats, or anything the agent should treat as project instructions. Gatekeep loads developer.md as a developer prompt.

Gatekeep also checks root-level markdown files such as README.md, CONTRIBUTING.md, or ARCHITECTURE.md and adds them to the model context by filename. The dashboard project table shows which markdown files were found and prompts you to add developer.md when it is missing.

Environment Variables

You can set config through .env (or UI settings):

AI_API_KEY=your_api_key
AI_BASE_URL=https://api.openai.com/v1
MODEL_NAME=gpt-4o

Local Data and Privacy

Gatekeep stores data under your OS config/app-support directory:

• config.json for provider/model/base URL/API key
• projects.sqlite for projects, chat history, activity traces, and usage

Nothing is uploaded except model API requests sent to your configured provider.

Common Commands

gatekeep --help
gatekeep start
gatekeep start --project .
pnpm run typecheck