Dimension 1: Agent count       → N agents in PTY sessions
Dimension 2: Agent direction   → bidirectional, mesh, meeting
Dimension 3: Agent depth       → recursive teams, fractal orchestration
Dimension 4: Agent machines    → cross-machine via gateway/P2P

Dimensions 1-4:  AI operates inside the terminal
Dimension 5:     AI sees and controls the operating system

  Terminal world:    files, code, git, shell commands
  OS world:          windows, buttons, apps, processes, screen, clipboard

See:     Windows, displays, sessions, and accessibility state
         Actionable targets and rendered text
         Processes, power, devices, ports, services, and security posture
         App data, logs, clipboard state, and visual evidence when needed

Do:      Act on a guarded target handle
         Press or set accessible controls by name
         Type, paste, send keys, and use pointer fallbacks
         Manage apps, windows, menus, and selected system settings
         Run the platform script surface when explicitly required

kernel-cli --version
kernel-cli --json doctor
kernel-cli --json report
kernel-cli --json targets --app "TextEdit" --intent save --top 20
kernel-cli --json act '<target_id>' --snapshot '<snapshot_id>' --dry-run

Raw (kernel-cli):          Harnessed (niia):
  Direct writes execute.      OTP required for control.
  Local audit is evidence.    Time-limited. Scope-limited.
  No approval gate.           Human holds the key.

# Observe through the read-only harness surface
niia observe targets --app "TextEdit" --json

# Unlock control through the human OTP flow
niia control unlock --scope full --duration 30m
niia control act '<target_id>' --snapshot '<snapshot_id>' --dry-run --json
niia control lock

App data / machine state
            ↓
Windows / accessibility / target handles
            ↓
OCR / screenshot / coordinate fallback

{
  "connector": "2.0",
  "name": "full-deploy",

  "pipeline": {
    "phases": [
      {
        "name": "implement",
        "model": "opus",
        "prompt": "Implement the feature. Commit and push.",
        "capabilities": { "pty": true }
      },
      {
        "name": "verify-ci",
        "model": "sonnet",
        "prompt": "Open Chrome. Navigate to GitHub PR. Check CI status. If green, click Merge.",
        "capabilities": {
          "pty": true,
          "kernel": {
            "observe": ["windows", "ax-tree"],
            "control": ["ax-press", "key"],
            "requires_otp": true
          }
        }
      },
      {
        "name": "notify",
        "model": "haiku",
        "prompt": "Open Slack. Send deploy notification to #engineering.",
        "capabilities": {
          "kernel": {
            "control": ["ax-press", "type", "key"],
            "requires_otp": true
          }
        }
      }
    ]
  }
}

Dimension 1: Agent count       "How many?"
Dimension 2: Agent direction   "Who talks to whom?"
Dimension 3: Agent depth       "How deep does the tree go?"
Dimension 4: Agent machines    "Which physical machines?"
Dimension 5: Agent OS access   "What can the agent see and touch?"

Each dimension is independent.
Each multiplies the others.
connector.json declares all five.

Dimensions 1-4:  AI that writes code
Dimension 5:     AI that uses the computer

Adding Dimension 5 to a pipeline means the AI can:
  - Write code (PTY)
  - Check CI status (browser via AX)
  - Merge the PR (button press via AX)
  - Notify the team (Slack via AX)
  - Monitor production dashboards (observe via AX)
  - Respond to alerts (control via AX)

This is not "AI coding assistant."
This is "AI computer user."

Dimensions 1-4:  Sandbox + worktree (software isolation)
Dimension 5:     Raw capability + an explicit human harness

Software isolation: AI can't write outside worktree
Harnessed control: NIIA refuses writes until the human completes OTP
Raw kernel-cli: direct writes are not OTP-gated

The more powerful the capability,
the stronger the harness.
The strongest harness is physical.