HELMStart
MCPLLMs

Start

Start

Install HELM, proxy one AI call, inspect a receipt, and verify the decision path.
PublicSource-ownedMarkdown export

Install HELM, run a protected AI framework, inspect receipts, and verify the decision path.

DiagramInstall CLI -> Pair Console -> Run protected framework -> Inspect signed receipts -> Verify EvidencePack
flowchart LR
  install["Install CLI"] --> pair["Pair Console"]
  pair --> launch["Run protected framework"]
  launch --> receipts["Inspect signed receipts"]
  receipts --> verify["Verify EvidencePack"]
Mermaid source
flowchart LR
  install["Install CLI"] --> pair["Pair Console"]
  pair --> launch["Run protected framework"]
  launch --> receipts["Inspect signed receipts"]
  receipts --> verify["Verify EvidencePack"]

Paths

Path Time What you get
Launchpad Quick Start ~5 min Run a real framework through HELM's execution boundary, get signed receipts and an EvidencePack
Kernel Boundary Demo ~10 min Start the local HELM server, exercise the PEP boundary, inspect and verify individual receipts

Which Path Should You Choose

Choose Launchpad if you want to protect a real AI agent framework (OpenClaw, Hermes, OpenCode, Kilo Code) and see end-to-end governed execution with Console integration.

Choose the Kernel Boundary Demo if you want to understand HELM's execution boundary at the API level — intercepting tool calls, issuing verdicts, and verifying receipts directly.

Both paths produce signed receipts and offline-verifiable EvidencePacks. Launchpad wraps the Kernel boundary in a higher-level launch lifecycle with sandbox isolation, policy packs, and Console dashboards.

Beyond the Kernel

The same boundary semantics scale up through the commercial surfaces:

Prerequisites

  • macOS with Homebrew
  • Docker Desktop (for Launchpad container substrates)
  • A Mindburn Console account at https://console.helm.mindburn.org

Source Truth

  • Launchpad reference: helm-ai-kernel/docs/LAUNCHPAD.md
  • Kernel quickstart: helm-ai-kernel/docs/QUICKSTART.md
  • CLI binary: helm-ai-kernel / helm
  • Install: brew install mindburnlabs/tap/helm-ai-kernel