Kiro-Go acts as a proxy service, converting Kiro accounts into APIs compatible with OpenAI and Anthropic standards. This project targets developers and power users needing to integrate Kiro's capabilities into existing LLM workflows, offering a unified endpoint for multi-account management, token refresh, and streaming responses, thereby simplifying integration and enhancing resource utilization.

How It Works

Kiro-Go functions as a reverse proxy, intercepting requests intended for OpenAI or Anthropic APIs and routing them to underlying Kiro accounts. It employs a round-robin load balancing strategy across a pool of configured Kiro accounts to distribute traffic. Key features include seamless automatic token refresh, support for streaming Server-Sent Events (SSE) responses, and a web-based administration panel for managing accounts and settings. The project also introduces an optional "thinking mode" that enables extended, step-by-step reasoning capabilities for LLM calls.

Quick Start & Requirements

The recommended installation method is via Docker Compose:

git clone https://github.com/Quorinex/Kiro-Go.git
cd Kiro-Go
mkdir -p data
docker-compose up -d

Alternatively, a docker run command is provided, requiring manual creation of a data directory for persistence. Building from source requires cloning the repository and using the Go toolchain (go build). A persistent data directory (mapped as a volume in Docker) stores configuration, including account details. The web admin panel is accessible at http://localhost:8080/admin.

Highlighted Details

• API Compatibility: Fully supports Anthropic's /v1/messages endpoint and OpenAI's /v1/chat/completions endpoint.
• Multi-Account Pooling: Implements round-robin load balancing across multiple Kiro accounts for efficient resource distribution.
• Authentication Flexibility: Supports various Kiro account authentication methods, including AWS Builder ID, IAM Identity Center (Enterprise SSO), SSO Token, Local Cache, and Credentials JSON import.
• "Thinking Mode": Extends LLM functionality by enabling step-by-step reasoning, configurable via a model name suffix and output format for both OpenAI and Claude APIs.
• Web Admin Panel: Provides a user-friendly interface for adding/managing accounts, configuring settings, and monitoring usage.
• Streaming & Usage Tracking: Offers real-time SSE responses and monitors requests, tokens, and credits.

Maintenance & Community

The project includes an automatic version update check, suggesting ongoing maintenance. However, the provided README does not detail specific contributors, sponsorships, or community channels like Discord or Slack.

Licensing & Compatibility

Kiro-Go is released under the MIT License, which is permissive and generally allows for commercial use and integration into closed-source projects.

Limitations & Caveats

This project is explicitly provided for educational and research purposes only and is not affiliated with Amazon, AWS, or Kiro. Users are solely responsible for ensuring their usage complies with all applicable terms of service and laws. The default administrator password must be changed before production deployment for security.