Summary ZyphrZero/kiro.rs is a Rust-based proxy server offering unified access to Anthropic Messages API and OpenAI Chat Completions API. It translates requests to Kiro or Amazon Q backends, enabling developers to manage multiple LLM credentials, automate token refreshing, and implement fault tolerance and observability for AI applications.

How It Works This Rust proxy intercepts API calls, translating standard OpenAI and Anthropic formats into Kiro/Amazon Q backend requests. It features sophisticated multi-credential management (IDC, Social, Enterprise SSO, Azure AD, Kiro API Keys) with automatic token refresh and persistence. Dynamic load balancing and failover mechanisms handle credential exhaustion, rate limiting, or token expiry. An integrated Web Admin panel provides centralized control.

Quick Start & Requirements Docker is the recommended deployment method via docker-compose.yml. Prerequisites: Docker. Setup involves creating a data directory, downloading docker-compose.yml, and running docker compose up -d. The API is available at http://<host>:8990/v1/messages, and the Admin UI at http://<host>:8990/admin. Pre-compiled binaries for major platforms are also available.

Highlighted Details

• Broad API Compatibility: Supports Anthropic Messages API (/v1/messages), OpenAI Chat Completions (/v1/chat/completions), and OpenAI Responses (/v1/responses), including synthesized SSE streams.
• Advanced Credential Handling: Manages diverse authentication types (Builder ID, IAM Identity Center, Social, Azure AD, Kiro API Key) with automatic token refresh, scheduling, and robust failover.
• Integrated Admin Panel: Web interface for managing credentials, client keys, proxy pools, viewing logs, and monitoring usage statistics and trace data.
• LLM Feature Support: Includes capabilities for tool use, reasoning/thinking, and image processing with automatic resizing and re-encoding.

Maintenance & Community A fork of hank9999/kiro.rs, community discussions are hosted on linux.do. The project is actively maintained with multi-platform releases and Docker images published on GitHub.

Licensing & Compatibility The README states "License 见 LICENSE" without specifying the license type. The project is explicitly for research and self-use ("本项目仅供研究和自用"), with commercial use implications not detailed.

Limitations & Caveats Designated for research and personal use, with no explicit commercial licensing. As a fork, its maintenance path may differ from its origin. The extensive feature set introduces complexity in configuration and management.