AaronWong1999/hermesclaw

Run Hermes Agent and OpenClaw on the same WeChat account

What it solves

HermesClaw solves the problem of running multiple AI agents on a single WeChat account. Normally, gateways for agents like Hermes Agent and OpenClaw exclusively lock the iLink connection, meaning only one can be active at a time; attempting to run both results in 403 errors and dropped messages. It also adds WeChat support to OpenCode, which does not have it natively.

How it works

It acts as a thin Python proxy and the sole iLink poller for the account. Instead of each agent connecting directly to the iLink API, HermesClaw runs local proxy servers and an ACP bridge. It routes incoming messages based on specific commands (like /hermes or /openclaw) and forwards raw iLink protocol messages to the respective agent gateways. This allows each gateway to believe it is talking to the real iLink API while sharing a single connection.

Who it’s for

Users who want to utilize multiple AI agents (Hermes Agent, OpenClaw, and OpenCode) simultaneously on one WeChat account to complement their capabilities, such as using one for general assistance and another for debugging or environment inspection.

Highlights

  • Multi-Agent Routing: Switch between agents or activate multiple agents at once using commands like /hermes, /openclaw, /opencode, /both, and /three.
  • Vibe Coding via WeChat: Enables OpenCode support on WeChat, allowing users to use voice messages for coding tasks.
  • Lightweight Architecture: A small Python proxy (~500 lines) that forwards raw messages without processing media or touching agent memory.
  • Smart Installation: Includes an automated installer that detects existing gateways, extracts tokens, and patches configurations automatically.

Related

  • Project
    onllm-dev/onWatch

    A free, open-source AI API quota monitoring tool that tracks usage and spending across multiple providers in a local dashboard to prevent unexpected throttling.

  • Project
    Storybloq/storybloq

    A project memory and workflow system for AI coding agents that persists stories, plans, and handovers in Git-tracked files to maintain context across sessions.

  • Project
    EleutherAI/sparsify

    A lean library for training and enable loading k-sparse autoencoders (SAEs) and transcoders on HuggingFace language model activations to aid in mechanistic interpretability.

  • Project
    symgraph/GhidrAssistMCP

    A Ghidra extension that implements the Model Context Protocol (MCP), enabling AI assistants to programmatically interact with Ghidra's reverse engineering capabilities.