Locals
Generated-Computers/Locals
Local model server for Macs built for agents, with a launcher that connects Hermes Agent and others
Locals is a local server for Macs built for agents rather than plain chat, with a locals launch command that connects Hermes Agent and other coding agents. It focuses on tool calls, long conversations and many agents running at once.
What Locals does
Locals runs AI models on your own Mac and is designed around agent traffic: tool calls that must be correct, prompts that grow each turn, and several agents working in parallel. The README claims a team of agents gets answers up to three times faster than the standard way of running models on a Mac, and it says nothing leaves your computer. Those figures are the authors' own and rest on a cited paper on grammar-checked inference.
Tool-call replies are written under a grammar applied on the GPU, so an agent can only call tools it was offered and arguments arrive in the type the schema asks for. locals code opens OpenCode in a project, and locals launch connects Hermes Agent, OpenClaw, Claude Code, Kilo Code, Cline, OpenHands and Aider. locals run splits a task among several agents, each working in its own copy of the project, and lets your tests decide. The first run downloads a model suited to your Mac, about 3 GB.
Key features
- Batches requests from many agents together on the Mac's GPU
- Grammar-constrained tool calls so replies parse
locals launchfor Hermes Agent, OpenClaw, Claude Code, Kilo Code, Cline, OpenHands and Aiderlocals runto split a task across agents in separate project copies- Works with apps that speak the OpenAI, Anthropic or Ollama APIs
When to use it
- Running Hermes Agent on a local model on a Mac with no cloud calls
- Running several coding agents on one machine without them queuing
- Testing a task split across agents with your test suite deciding
Who it is for: Mac users who want local models to serve Hermes Agent and other coding agents.
How it fits with Hermes Agent
Locals is a general local server, and Hermes Agent is one of the agents locals launch can connect.
How to install Locals
These commands are copied from the project's README. Check the repository for the latest steps before you run them.
curl -LsSf https://raw.githubusercontent.com/Generated-Computers/Locals/main/install.sh | sh
uv tool install --python 3.12 localsRequirements: A Mac (the README targets Apple Silicon); about 3 GB is downloaded for the first model
FAQ
What is Locals?
Locals is a local model server for Macs built for agents, handling tool calls, long conversations and many agents working at once.
Does Locals work with Hermes Agent?
Yes. The README says locals launch connects Hermes Agent, as well as OpenClaw, Claude Code, Kilo Code, Cline, OpenHands and Aider.
How do I install Locals?
Run the install.sh script from the repository with curl, or use uv tool install --python 3.12 locals to install from PyPI. Then run locals code in a project folder.
Similar models for Hermes Agent
All modelsAI middleware for model routing, RAG and OpenAI-compatible APIs, with Hermes Agent sync hooks
humantonylee free-routerTerminal tool that pings free AI models and writes the best one into OpenCode, OpenClaw or Hermes Agent
soulduse AI Token MonitorMenu bar and system tray app that prices your AI coding tool usage from local logs, Hermes Agent included
y13sint FreeQwenApiLocal OpenAI-compatible proxy that exposes a Qwen Chat web account as an API for tools such as Hermes Agent
EfficientContext ContextPilotContext reuse layer that cuts redundant tokens before they reach the model
dyedd LensSelf-hosted multi-protocol LLM gateway with routing, failover and request logs, usable with Hermes
Related guides: How to install Hermes Agent