yinnho

AginxBrain — AI skill for Claude Code

AI community

Local-first AI gateway with bidirectional protocol conversion (Anthropic ↔ OpenAI ↔ Responses).

How to install AginxBrain

This entry records only its repository, not the path inside it, so there is no exact command to give. Open yinnho/AginxBrain and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AginxBrain does

Local-first AI gateway with bidirectional protocol conversion (Anthropic ↔ OpenAI ↔ Responses). Route Claude Code / Codex CLI to any provider by quality tag, with failover. Rust + Tauri.

Alternatives in AI

Axonhub — ⚡️ Open-source AI Gateway — Use any SDK to call 100+ LLMs 3.2k ★
Sentrux — Real-time architectural sensor that helps AI agents close the feedback loop, enabling recursive self-improveme 2.9k ★
My Free Code — Open-source multi-provider AI gateway for Claude Code and other coding agents, with model routing, streaming 633 ★

README

AginxBrain

**Open-source AI gateway · Bidirectional protocol conversion · Tag-based routing**

Speak Anthropic, OpenAI, or Responses on the client side — any of them on the provider side. AginxBrain converts between them, routes by tag, and fails over automatically.

[![Version](https://img.shields.io/badge/version-0.4.0-blue)](https://github.com/yinnho/AginxBrain/releases) [![License: MIT](https://img.shields.io/badge/license-MIT-green)](./LICENSE) [![Platform](https://img.shields.io/badge/platform-macOS%20%7C%20Windows%20%7C%20Linux%20%7C%20Server-lightgrey)](https://github.com/yinnho/AginxBrain/releases) [![Rust](https://img.shields.io/badge/built%20with-Rust-orange)](https://www.rust-lang.org/)

[Features](#features) · [Architecture](#architecture) · [Quick Start](#quick-start) · [Configuration](#configuration) · [Comparison](#comparison)

AginxBrain sits between your AI client (Claude Code, Codex CLI, any OpenAI / Anthropic / Responses SDK) and your model providers, and does four things transparently:

Protocol conversion — clients speak Anthropic Messages, OpenAI Chat, or OpenAI Responses; providers speak any of the three. All 9 combinations, both directions, streaming included.
Tag-based routing — call a model by its tag name (e.g. model: "sonnet", model: "gpt-5.5") and AginxBrain maps it to the right provider and model, with multi-candidate failover chains.
Reliability — per-route circuit breaker, automatic failover on retryable errors, smart auto-routing that upgrades the tier based on the request, hot-reloaded config.
Key custody & observability — providers' real keys live in one place behind per-caller API keys; every request is logged with token usage and estimated cost.

It runs in two shapes:

Server (the gateway) — a single self-hosted binary. Holds all the keys, runs the proxy, embeds an admin dashboard at http://host:port/.
Desktop (the client) — a thin