KiroGate is an API proxy gateway designed to make OpenAI and Anthropic APIs compatible with tools that support Kiro IDE, particularly Claude Code CLI. It targets developers needing a unified interface for LLM interactions, offering enhanced features beyond standard API access. The primary benefit is enabling broader tool integration and adding advanced functionalities like image processing and user management.

How It Works

This project acts as a FastAPI-based reverse proxy, translating requests to be compatible with both OpenAI and Anthropic API specifications. It supports image inputs in both OpenAI (image_url) and Anthropic (image) formats, integrates Anthropic's web_search tool via its MCP API, and allows configuration of HTTP/SOCKS5 proxies for API access. Key architectural choices include robust token management, automatic credential refreshing, and security hardening against common vulnerabilities.

Quick Start & Requirements

• Primary Install/Run: Docker is recommended: docker run -d -p 8000:8000 -e PROXY_API_KEY="..." -e ADMIN_PASSWORD="..." -e ADMIN_SECRET_KEY="..." -v kirogate_data:/app/data ghcr.io/awei84/kirogate:main. Alternatively, clone the repo, install dependencies (pip install -r requirements.txt), configure .env, and run python main.py.
• Prerequisites: Python 3.10+, a Kiro IDE login, and potentially a Kiro refresh token (obtainable via Kiro Account Manager or manual packet capture).
• Links: GitHub Repository

Highlighted Details

• Supports IDC (Builder ID) authentication.
• Integrates Anthropic's web_search tool.
• Handles image inputs for both OpenAI and Anthropic API formats.
• Features a user system with LinuxDo/GitHub OAuth2 login, token donation, and API key generation.
• Includes an Admin management backend for user, token, and IP blacklisting.
• Provides OpenAI and Anthropic compatible APIs with streaming, automatic retries, and tool calling capabilities.
• Offers enhanced token management with refresh anti-jitter and account detail viewing.

Maintenance & Community

This project is a fork of kiro-openai-gateway by @Jwadow. The README does not specify active maintainers, community channels (like Discord/Slack), or a public roadmap.

Licensing & Compatibility

• License Type: GNU Affero General Public License v3.0 (AGPL v3.0).
• Compatibility Notes: AGPL-3.0 is a strong copyleft license. Its terms require that any modified versions of the software, when made available to users over a network, must also be released under the AGPL-3.0 license, including the source code. This can impose significant obligations for commercial use or integration into proprietary systems.

Limitations & Caveats

The AGPL-3.0 license's network-based copyleft provisions are the primary adoption consideration, potentially requiring open-sourcing of derivative works if deployed as a service. Obtaining Kiro refresh tokens may require manual steps or specific tooling. The project's reliance on external Kiro IDE services means its availability is tied to those services.