Bizra Node0 Proactive Pilot

Peak Masterpiece Protocol

Execute every non-trivial operational decision through this 6-phase loop. Phases map to Boyd's OODA (Observe-Orient-Decide-Act) extended with Deming's PDCA quality cycle and Besta's Graph-of-Thoughts reasoning.

Phase 1 — Observe

Gather runtime state through non-mutating commands:

• python scripts/node0_activate.py status
• Preflight checks (repo root, Python/venv, LM Studio, token, PID/log state)
• Recent smoke results if available

Collect raw evidence before forming any hypothesis.

Phase 2 — Orient (GoT Expansion)

Diverge into ≥3 hypotheses for the operational situation. Apply the Graph-of-Thoughts workflow defined in references/got-execution-graph.md:

1. Generate candidate interpretations of the observed state.
2. Cross-validate each hypothesis against runtime evidence.
3. Flag contradictions between hypotheses explicitly.

Do not collapse to a single recommendation prematurely.

Phase 3 — Synthesize (Interdisciplinary Convergence)

Evaluate the surviving hypotheses through the 7-lens matrix defined in references/interdisciplinary-lens-matrix.md:

• Systems · Reliability · Security · Economics · Ethics · Operations · Product Impact

Each lens can veto a hypothesis if its veto condition is triggered. Only hypotheses that survive all 7 lenses advance.

Phase 4 — Score (SNR/Ihsān Rubric)

Score the converged recommendation using the rubric in references/snr-ihsan-scorecard.md:

Apply tier gates:

Source: core/integration/constants.py — UNIFIED_SNR_THRESHOLD (0.85), SNR_THRESHOLD_T1_HIGH (0.95), SNR_THRESHOLD_T0_ELITE (0.98), STRICT_IHSAN_THRESHOLD (0.99).

Phase 5 — Gate (Constitutional Checks)

Apply constitutional filters before any output:

• Ihsān ≥ 0.95 for production operations (UNIFIED_IHSAN_THRESHOLD)
• Daughter Test: "Would I be comfortable if my daughter were affected by this action?"
• ConstitutionalGate pattern: APPROVED / NEEDS_REVIEW / REJECTED

Reference: core/governance/constitutional_gate.py, docs/PROACTIVE_SOVEREIGN_ENTITY.md § Constitutional Filters.

Phase 6 — Execute/Advise (Human-in-Loop)

Present the scored, gated recommendation with:

• Exact command to run
• Expected artifacts and success criteria
• Giants provenance (≥2 giants + ≥1 repo artifact path)
• Confidence tier label

For mutating operations: stop and await explicit user confirmation before execution.

Standing on Giants Protocol

Every non-trivial recommendation MUST include provenance:

• Minimum: 2 giants + 1 concrete repo artifact path
• Format: Standing on Giants: Name (contribution) · Name (contribution) — artifact: path/to/file

Canonical giants for this skill:

Intent Router

Map user intent to one command path and report expected artifacts.

Guarded Preflight (Mandatory Before Mutating Ops)

Run all checks before any mutating operation:

1. Repo root and entrypoints:
   • pwd — confirm BIZRA-DATA-LAKE root
   • test -f scripts/node0_activate.py
   • test -f scripts/start_proactive.sh
   • test -f scripts/stop_proactive.sh
2. Python and venv:
   • python --version — require 3.11+
   • test -d .venv-linux
3. LM Studio reachability:
   • curl -sS -m 5 http://192.168.56.1:1234/v1/models
4. Token presence (dual-token awareness):
   • test -n "$LM_API_TOKEN" — preferred for node0_activate.py
   • Note: some subsystems use LM_STUDIO_API_KEY (set via scripts/set_lm_studio_key.sh). If LM_API_TOKEN is unset but LM_STUDIO_API_KEY is set, advise the user to export LM_API_TOKEN explicitly.
5. PID and log state:
   • test -f sovereign_state/proactive.pid && cat sovereign_state/proactive.pid || true
   • ls -la logs/proactive 2>/dev/null || true

If any preflight check fails, switch to diagnostics-only mode and do not run mutating commands.

Explicit confirmation required immediately before:

• python scripts/node0_activate.py start
• ./scripts/start_proactive.sh ...
• ./scripts/stop_proactive.sh
• python scripts/node0_activate.py mission "..."

SNR/Ihsān Gate Policy

Tiered quality gates aligned with authoritative constants from core/integration/constants.py:

All thresholds are imported from core/integration/constants.py. Do NOT hardcode alternative values.

Operational Procedures

Status

python scripts/node0_activate.py status

Report: LM Studio connectivity, loaded model count, token state, configured proactive mode.

Start (Path A — Foreground)

Use for direct Node0 activation. Does NOT create PID file or log artifacts — runs in foreground with signal handling.

python scripts/node0_activate.py start

Start (Path B — Daemon)

Use for background daemon with explicit mode/config control. Creates PID and log artifacts.

./scripts/start_proactive.sh --mode <mode> --config config/proactive_config.yaml

Supported modes: reactive, proactive_suggest, proactive_auto, proactive_partner. Artifacts: sovereign_state/proactive.pid, logs/proactive/sovereign.log, logs/proactive/startup.log.

Stop

Graceful stop first (SIGTERM, 30s timeout):

./scripts/stop_proactive.sh

Force-stop only after graceful path exhaustion:

./scripts/stop_proactive.sh --force

Mission

In-process mission execution — does NOT require background daemon. Requires LM Studio reachable.

python scripts/node0_activate.py mission "<task>"

Report: assigned PAT agents, per-agent result, total tokens, Ihsān score.

Validation Procedures

Run autonomous pilot smoke suite:

pytest tests/integration/test_autonomous_pilot.py -q

8 Smoke Pillars

• Full pass → Node0 pilot baseline operational.
• Any failure → identify failing pillar, stop mutating ops, enter triage via references/failure-triage.md.

Response Contract v2

Every proactive pilot response MUST include these sections in order:

1. Node/Runtime Status — running / stopped / degraded, with evidence command
2. Active Mode — current proactive mode (reactive | proactive_suggest | proactive_auto | proactive_partner)
3. Interdisciplinary Lens Synthesis — which lenses were evaluated, any vetoes triggered
4. GoT Path Summary — hypotheses considered, convergence rationale
5. SNR/Ihsān Score Tier — computed tier label and numeric score
6. Blockers — any preflight failures, degraded subsystems, or veto conditions
7. Safest Next Action — single unambiguous next command or diagnostic step
8. Giants Provenance — ≥2 giants cited with ≥1 repo artifact path

For simple status checks, sections 3–5 may be abbreviated. For mutating operations, all 8 sections are mandatory.

Escalation And Safe Fallbacks

When blocked:

1. Switch to diagnostics-only flow — no mutating operations.
2. Report exact failing preflight check and impacted command path.
3. Propose minimal safe remediation from references/failure-triage.md.
4. Re-run preflight before any retry.

Escalate when repeated failures persist (≥3 consecutive):

1. Collect command outputs and log snippets from logs/proactive/.
2. Run targeted triage from references/failure-triage.md.
3. Freeze mutating operations.
4. Direct user to docs/PROACTIVE_SOVEREIGN_ENTITY.md and deploy/node0/README.md for manual intervention.

References

• references/command-map.md — canonical command matrix and source file mapping
• references/failure-triage.md — symptom-driven diagnosis and remediation
• references/interdisciplinary-lens-matrix.md — 7-lens evaluation framework
• references/got-execution-graph.md — Graph-of-Thoughts decision workflow
• references/snr-ihsan-scorecard.md — scoring rubric and tier definitions