pleaseai/shunt

Shunt Claude Code agents to any model — selective, per-agent inference-layer routing proxy

What it solves

shunt acts as a transparent gateway that allows you to use different LLM providers (like OpenAI, xAI, or Gemini) with tools designed specifically for Claude Code. It solves the problem of being locked into a single model provider while wanting to maintain the functionality of your existing coding agent workflow.

How it works

It functions as a proxy server that sits between your coding agent and the LLM provider. When your agent sends a request, shunt intercepts it at the inference layer. Based on your configuration, it routes the request to a different backend provider while keeping the agent's tools and skills intact. It supports various providers by either passing through Anthropic-compatible requests or translating them into the format required by other services (like OpenAI's Responses API).

Who it’s for

This tool is for developers who use Claude Code or similar coding agents and want the flexibility to swap out the underlying model for alternatives like GPT-4, Grok, or others, without needing to change their development environment or agent configuration.

Highlights

  • Model Routing: Diverts specific model IDs to different providers while leaving others as-is.
  • Broad Provider Support: Built-in support for OpenAI, xAI (Grok), Cursor, Gemini, Kimi, and other Anthropic-compatible backends.
  • Subscription Reuse: Allows you to leverage existing subscriptions for various AI services through a unified interface.
  • Advanced Features: Includes support for multi-account pooling, failover logic, and an optional admin dashboard for monitoring usage.
  • Transparent Integration: Operates as a local service that integrates seamlessly with existing CLI tools via environment variables.

Related

  • Project
    pingmike2/freebuff2api-wokers

    A single-file API proxy that turns freebuff's free models into an OpenAI-compatible API, with a web panel for managing multiple accounts, quotas, and session rotation.

  • Project
    LeiWang1999/ZYNQ-NVDLA

    ZYNQ‑NVDLA is an open‑source project that ports NVIDIA’s Deep Learning Accelerator (NVDLA) to Xilinx Zynq FPGA boards. It provides RTL, Vivado IP, Linux drivers, pre‑built ARM64 libraries, and a Tengine‑Lite backend so you can run quantised neural‑network models (e.g., ResNet‑18, YOLOX‑nano) on the FPGA‑based accelerator with low latency.

  • Project
    Leavesfly/TinyAI

    A full-stack AI framework built entirely in Java that provides a complete pipeline from low-level tensor computation to LLMs, AI agents, and embodied intelligence.

  • Project
    MiroPsota/torch_packages_builder

    A toolset for automating the build and indexing of PyTorch-based packages with custom operators, providing pre-built wheels via a PEP 503 compliant index.