aimod-cc/agent2api

What it solves

Agent2API transforms the login states of various AI desktop clients into a single, local OpenAI-compatible API gateway. This allows users to use their existing account quotas from multiple AI providers without needing official API keys or modifying their clients, provided the client supports a custom base_url.

How it works

The project acts as a local reverse proxy. It captures the session/login state of supported AI platforms and forwards requests to the official upstream gateways. It provides a unified base_url (http://127.0.0.1:3065/v1) and translates various proprietary wire protocols into standard OpenAI-compatible chat completions, responses, and messages formats.

Who it’s for

Users who have accounts with multiple AI desktop clients and want to centralize their access through a single API endpoint for use in their own tools or OpenAI-compatible software.

Highlights

• Broad Provider Support: Supports a wide range of platforms including WorkBuddy, Trae, Cline, Qoder, and others, with varying levels of support for LLM requests, token renewal, and balance queries.
• Unified API: Exposes a OpenAI-compatible interface for /v1/chat/completions, /v1/responses, and /v1/models across all platforms.
• Account & Model Management: Includes a GUI (Tauri-based desktop app or Docker-based web panel) for managing accounts, priorities, and model mapping.
• Advanced Routing: Features a global priority queue, 429 (rate limit) degradation, and outbound fingerprint desensitization to avoid detection.
• Operational Tools: Provides request reports with success rates and token usage, as well as automated tasks for daily check-ins and credential maintenance.
• Deployment Flexibility: Available as a standalone Windows installer or a Docker container for headless deployment.