Helm

HELM — Operator Shell For Local AI Systems

HELM is a desktop ops shell for developers running AI agents, local services, and multi-machine model workflows. It keeps overall system status visible while giving each domain its own page instead of forcing everything into one dense dashboard.

Current branch version: `4.0.2`. Latest tagged release: `v4.0.2`.

The 4.0 line turned HELM into an operator shell with page ownership instead of panel sprawl. Recent releases added Swarm drill-down, persistent local history, a real traffic-topology grid, and a mapped radio signal globe that traces station origin back to an operator-defined home endpoint. Current mainline also carries Registry and Sentinel work targeting the next line.

[!IMPORTANT] If a clean clone breaks on first run, do not edit source paths by hand. Fix it one of two ways:

• Guided setup: paste `docs/llm-setup-prompt.md` into ChatGPT, Claude, or any other LLM and follow it from the repo root.
• Agentic repair: hand `CLAUDE.md` to your coding agent, tell it to work from the repo root, copy .env.example to .env, and repair setup through config or env overrides only.

Quick Start (2 minutes)

New here? Paste `docs/llm-setup-prompt.md` into your AI assistant for guided setup.

git clone https://github.com/kunalnano/hydra.git
cd hydra
cp .env.example .env
npm install
npm run dev

Requires Node.js 18+, LM Studio running locally, and at least one model loaded in LM Studio. The repo is still named `hydra`, but the desktop app window and UI are branded `HELM`.

What's New

• Removed repo-root and machine-layout assumptions from startup by resolving .env, registry seed data, Sentinel config, HIVE role files, and helper scripts relative to the current clone or packaged resources
• Replaced developer-specific defaults with documented .env overrides for LM Studio, repo scanning, external agent feeds, log paths, Sentinel outputs, HIVE paths, and