<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> This project provides a proxy API gateway that enables users to access Claude models, typically via AWS CodeWhisperer/Kiro, through standard OpenAI or Anthropic compatible clients. It targets developers and users seeking to integrate powerful LLMs like Claude into various applications and workflows using familiar API interfaces, abstracting away direct Kiro/AWS integration complexities.

How It Works

The gateway acts as an intermediary, translating requests from OpenAI or Anthropic compatible formats into the Kiro API format. It automatically handles authentication using Kiro IDE credentials, Kiro CLI AWS SSO, or direct refresh tokens. Key features include support for vision, tool calling, streaming, and automatic retry logic, abstracting away the underlying Kiro API specifics and providing a unified interface.

Quick Start & Requirements

• Installation: Clone the repository or download as ZIP, then pip install -r requirements.txt.
• Prerequisites: Python 3.10+, Git. Requires a logged-in Kiro IDE or Kiro CLI with AWS SSO (Builder ID).
• Configuration: Set PROXY_API_KEY and provide Kiro credentials via .env file, JSON credentials file, or kiro-cli SQLite database.
• Run: python main.py (or python main.py --port 9000). Server available at http://localhost:8000.
• Docs: Multiple language READMEs are available at the top of the repository.

Highlighted Details

• Offers both OpenAI-compatible (/v1/chat/completions) and Anthropic-compatible (/v1/messages) API endpoints.
• Supports advanced features like Vision, Tool Calling, and SSE Streaming.
• Includes automatic retry logic for common API errors (403, 429, 5xx) and smart token management.
• Provides "Extended Thinking" capabilities and smart model resolution for various Claude versions.

Maintenance & Community

The project is maintained by @Jwadow. Sponsorship links are provided for supporting development. No explicit community links (like Discord/Slack) are listed in the README.

Licensing & Compatibility

Licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). This license permits commercial use but requires that any modifications or derivative works, when distributed or made available over a network, must also be released under the AGPL-3.0. This copyleft provision is a significant consideration for commercial adoption.

Limitations & Caveats

Model availability is dependent on the user's Kiro tier (free/paid). The AGPL-3.0 license imposes strong copyleft requirements, mandating source code disclosure for network-accessible modified versions, which may restrict use in some proprietary commercial contexts. The project is not officially affiliated with AWS, Anthropic, or Kiro.