━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
                    vssh — AI Execution Runtime
        "SSH was designed for humans. vssh was designed for AI agents."
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  Show me my fleet                                                           │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh ─────────────────────────────────────────────────────────────╮
│                                                                             │
│  ┌──────────┬────────┬───────────────┬───────────────────┬──────┬────────┐  │
│  │   Name   │ Status │     Tags      │   Capabilities    │ Mem  │ Uptime │  │
│  ├──────────┼────────┼───────────────┼───────────────────┼──────┼────────┤  │
│  │ gpu-01   │ 🟢     │ ai-workload   │ cuda, gpu, ollama │  13% │ 32d    │  │
│  │ gpu-02   │ 🟢     │ docker-host   │ cuda, gpu, ollama │  18% │ 10d    │  │
│  │ gpu-03   │ 🟢     │ docker-host   │ cuda, gpu, ollama │  15% │ 36d    │  │
│  │ gpu-04   │ 🟢     │ docker-host   │ cuda, gpu, ollama │  56% │ 36d    │  │
│  │ web-01   │ 🟢     │ vps, nginx    │ linux, server     │   7% │ 31d    │  │
│  │ web-02   │ 🟢     │ vps, nginx    │ linux, server     │  24% │ 145d   │  │
│  │ db-01    │ 🟢     │ database      │ postgres          │  45% │ 108d   │  │
│  │ storage  │ ⚪     │ nas           │ nfs, backup       │  38% │ stale  │  │
│  └──────────┴────────┴───────────────┴───────────────────┴──────┴────────┘  │
│                                                                             │
│  8 hosts: 7 online, 1 stale                                                 │
│  4 GPU nodes (cuda/ollama), 2 web servers, 1 database, 1 NAS               │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  Check GPU usage                                                            │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh ─────────────────────────────────────────────────────────────╮
│                                                                             │
│  ┌────────┬───────────────────────┬─────────────┬──────┬──────┐             │
│  │  Host  │          GPU          │    VRAM     │ Temp │ Util │             │
│  ├────────┼───────────────────────┼─────────────┼──────┼──────┤             │
│  │ gpu-01 │ 2× NVIDIA RTX A6000   │ 33GB / 98GB │ 40°C │   0% │             │
│  │ gpu-02 │ 1× NVIDIA RTX 4090    │ 13GB / 24GB │ 56°C │  71% │  ← active   │
│  │ gpu-03 │ 1× NVIDIA GB10        │ (unified)   │ 57°C │  58% │  ← active   │
│  │ gpu-04 │ 1× NVIDIA GB10        │ (unified)   │ 47°C │   0% │             │
│  └────────┴───────────────────────┴─────────────┴──────┴──────┘             │
│                                                                             │
│  5 GPUs total • 2 actively running inference • 155GB VRAM available         │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  What's running on gpu-02?                                                  │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh (Fleet Memory) ──────────────────────────────────────────────╮
│                                                                             │
│  gpu-02:                                                                    │
│    Role: Production inference server                                        │
│    Services: ollama, docker                                                 │
│    Current model: llama3.1:70b (loaded, 13GB VRAM)                         │
│                                                                             │
│    Recent notes:                                                            │
│    • [Jun 25] Upgraded to RTX 4090, migrated from 3090                     │
│    • [Jun 20] Deployed llama3.1:70b for customer chatbot                   │
│    • [Jun 15] Docker cleanup, freed 200GB disk                             │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  Restart nginx on all web servers                                           │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh (Intent → Policy → Execution) ───────────────────────────────╮
│                                                                             │
│  ⚠️  This will restart nginx on 2 hosts. Proceed? [y/N]                     │
│                                                                             │
│  Plan:                                                                      │
│  ┌─────────┬─────────────────────────┬────────────┐                         │
│  │  Host   │        Command          │   Policy   │                         │
│  ├─────────┼─────────────────────────┼────────────┤                         │
│  │ web-01  │ systemctl restart nginx │ ✓ allowed  │                         │
│  │ web-02  │ systemctl restart nginx │ ✓ allowed  │                         │
│  └─────────┴─────────────────────────┴────────────┘                         │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ User ──────────────────────────────────────────────────────────────────────╮
│  yes                                                                        │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh (Structured Evidence) ───────────────────────────────────────╮
│                                                                             │
│  ✓ Completed on 2 hosts                                                     │
│                                                                             │
│  ┌─────────┬──────┬────────────┬───────────┐                                │
│  │  Host   │ Exit │  Duration  │ Transport │                                │
│  ├─────────┼──────┼────────────┼───────────┤                                │
│  │ web-01  │  0   │    142ms   │ TLS 1.3   │                                │
│  │ web-02  │  0   │    138ms   │ TLS 1.3   │                                │
│  └─────────┴──────┴────────────┴───────────┘                                │
│                                                                             │
│  Audit: action logged with hash a3f2c1... (key: admin@controller)          │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  Delete all logs on gpu-01                                                  │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh (Policy Block) ──────────────────────────────────────────────╮
│                                                                             │
│  🚫 Blocked by policy                                                       │
│                                                                             │
│  Command: rm -rf /var/log/*                                                 │
│  Reason:  Destructive command requires explicit approval                    │
│                                                                             │
│  To proceed, the operator must set allow_dangerous: true                    │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  What changed on db-01 this week?                                           │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh (Audit Diff) ────────────────────────────────────────────────╮
│                                                                             │
│  db-01 changes (Jun 21 - Jun 27):                                           │
│                                                                             │
│  ┌────────────┬──────────────────────────────────┬─────────────────┐        │
│  │    Date    │            Action                │      Key        │        │
│  ├────────────┼──────────────────────────────────┼─────────────────┤        │
│  │ Jun 27     │ pg_dump > backup.sql             │ backup@cron     │        │
│  │ Jun 25     │ psql: ALTER TABLE users ADD...   │ admin@laptop    │        │
│  │ Jun 23     │ systemctl restart postgresql     │ deploy@ci       │        │
│  │ Jun 21     │ apt upgrade postgresql-16        │ admin@laptop    │        │
│  └────────────┴──────────────────────────────────┴─────────────────┘        │
│                                                                             │
│  4 actions by 3 keys • 0 policy violations • 0 dangerous blocked            │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


╭─ User ──────────────────────────────────────────────────────────────────────╮
│  Deploy the new model to gpu-03                                             │
╰─────────────────────────────────────────────────────────────────────────────╯

╭─ Claude + vssh (Workflow) ──────────────────────────────────────────────────╮
│                                                                             │
│  Running workflow: deploy-ollama-model                                      │
│                                                                             │
│  [1/4] ✓ Stop current model                          (gpu-03)    0.8s      │
│  [2/4] ✓ Pull llama3.1:70b                           (gpu-03)   45.2s      │
│  [3/4] ✓ Start model with 8-bit quantization         (gpu-03)    3.1s      │
│  [4/4] ✓ Health check passed                         (gpu-03)    0.5s      │
│                                                                             │
│  ✅ Workflow completed in 49.6s                                             │
│                                                                             │
│  Fleet Memory updated:                                                      │
│    gpu-03.services = ["ollama", "docker"]                                   │
│    gpu-03.notes += "Deployed llama3.1:70b (Jun 27)"                        │
│                                                                             │
╰─────────────────────────────────────────────────────────────────────────────╯


━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

  vssh: The AI Execution Runtime

  ✓ Fleet Memory    — AI remembers your infrastructure
  ✓ Intent          — Natural language → verified command plans
  ✓ Workflow        — Multi-step operations, one command
  ✓ Policy          — Scoped access, dangerous commands blocked
  ✓ Evidence        — Structured results, not text to parse
  ✓ Audit           — Hash-chained, attributed, tamper-evident

  github.com/zeus-kim/vssh

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
