• title: HermQube — Primary Agent Runtimetype: entitycreated: 2026-06-10tags: [hermqube, agent, runtime, qubes, primary]
• HermQube — Primary Agent Runtime
  • What Is HermQube?
  • Hardware
  • Security Model
  • Software Stack
  • Running Services
  • Network
  • Constraints
  • Path to Full Operation

§title: HermQube — Primary Agent Runtime type: entity created: 2026-06-10 tags: [hermqube, agent, runtime, qubes, primary]

§HermQube — Primary Agent Runtime

§What Is HermQube?

HermQube is the primary agent runtime environment — a QubesOS AppVM that serves as the coordination hub for all Kapnet operations. It runs OWL (this instance) and coordinates with MKCTP-Alpha agents on macOS.

§Hardware

• CPU: Intel Core i5-6500 @ 3.20GHz (2 cores, no HT)
• RAM: 3.8GB total / ~3.1GB available
• Storage: 1TB Samsung SSD (shared-ro/rw/private partitions) + 16GB xvdb
• Network: Qubes NAT (10.137.0.2/32, gateway 10.138.26.110)
• GPU: None (headless, terminal-only)

§Security Model

• QubesOS: Compartmentalized security domains
• No C compiler: Cannot build Rust locally (cross-compile on Mac Mini required)
• SSD-encrypted: All persistent data on encrypted SSD
• Airgap-capable: SanDisk backup for offline operations

§Software Stack

• Node.js v22.22.3: Primary runtime for bridges and scripts
• Python 3.13.5: PDF generation, curriculum, analysis
• Git: Version control for wiki and configs
• nostr-tools (npm): Nostr event publishing
• pdfkit (npm): PDF generation
• kapnet-source (Rust): Full protocol implementation (not yet buildable here)

§Running Services

§Network

• Nostr relays: wss://relay.damus.io, wss://nos.lol
• Web reader: http://localhost:8080/reader.html
• kapnetd IPC: ~/.kapnet/kapnet.sock (not running — no C compiler)

§Constraints

1. No C compiler: Cannot build Rust. Mac Mini required for kapnetd.
2. Limited RAM: 3.8GB total. No local LLM possible.
3. Qubes NAT: No direct inbound connections. All outbound via relay.
4. No GPU: No ML acceleration. CPU-only inference only.

§Path to Full Operation

The Mac Mini M4 Pro resolves all constraints:

• 24GB unified RAM → local LLM inference
• Rust toolchain → build and run kapnetd
• Self-hosted relay → reduced public relay dependence
• Docker → containerized services
• Web gateway → public-facing Kapnet interface