CLARA Kill Web OODA

15-Namespace DAG -- Closed-Loop OODA with BDA Feedback Documentation
Scenario System

Select OODA Scenario

OODA-expanded pipeline: 105 rules across 15 namespaces with closed-loop Observe-Orient-Decide-Act-Assess engagement cycles and BDA feedback. Includes the full doctrine chain plus bda_assessment and loop_control namespaces.

Tactical Situation

Tactical Map

● Armor ▲ Air Defense ■ Soft Target ◆ Civilian ★ Platform ☆ Protected Asset
NSL entities: + Medical ☪ Religious ☐ Educational ☖ Civilian

ROE Status

NSL Entities (0)

Targets (0)

Platforms (0)

OODA Engagement Pipeline

Mission Profile OBJECTIVE

Select a mission profile to apply objective rules that shape engagement scoring without modifying constraint rules.

ML Model CLASSIFIER

Composite (recommended) -- runs both CNN and LR, composes predictions through AR-governed fusion. Agreement boosts confidence; disagreement triggers more conservative decisions. Same zero-violation guarantee.

OODA Loop Configuration CLOSED-LOOP

OODA-Aware Decisions (Final Cycle)

Run the OODA loop to see engagement decisions with BDA-informed re-evaluation.

15-Namespace DAG Inspector

DAG Topology

Features flow through 15 rule namespaces arranged in 5 layers. Includes bda_assessment and loop_control namespaces for closed-loop OODA.

Features (40-D)

Target Assessment

Weapons Pairing

ROE Compliance

Tactical Priority

No-Strike List

Collateral Objects

CDE L1

CDE L2

CDE L3

CDE L4

CDE L5

Eng. Authority

LOAC

BDA Assessment

Loop Control

BFS Trace

Select a target to see the OODA BFS trace.

DAG vs Flat Comparison

Select a target to compare DAG-composed vs flat scores.

Verification & Decision Trace

ErgoAI Formal Verification (Flora-2 / XSB)

ErgoAI re-derives the engagement decision independently in Flora-2 using final_decision/2 over the rule firings, and uses dominant_blocking_namespace/2 with \overrides/2 to identify which precedence-class blocker dominates when multiple namespaces fire. Python composes the same firings as a cross-check; any divergence is reported.

Rule Consistency

Runs contradictory_same_namespace/3, mutually_exclusive_pair/4, and redundant_pair/3 queries against the rule meta-data.

Constraint Invariance & Non-Interference

Queries \overrides/2 against the 4101 facts in precedence.ergo to confirm objective namespaces cannot override constraint namespaces.

Per-Target Formal Proof

Asserts the target's features into ErgoAI and queries final_decision/2, final_score/2, final_cap/2, and dominant_blocking_namespace/2. The result is checked against the OODA pipeline's decision.

Sample Proof Certificate

A canonical proof certificate produced by flora2_final_decision/2 on a representative engage target.

Trace Decision Trace

Pipeline audit trail showing which doctrine checks each decision traversed. Not an independent proof -- see ErgoAI above for formal verification.

Rules

All doctrine rules (90+ from published sources) plus OODA-specific rules in bda_assessment and loop_control namespaces. OODA rules are highlighted with purple accents.

Doctrine Rule Precedence Hierarchy

Higher-precedence rules cannot be overridden by lower levels. User rules (Level 6) can only ADD restrictions.

Level 1: LOAC (immutable -- international law, non-derogable)
Level 2: NSL (overridable: dual-use + commander auth only)
Level 3: ROE (commander adjustable within LOAC bounds)
Level 4: CDE (procedural, thresholds adjustable)
Level 5: Tac (fully adjustable by commander)
Level 6: User (additive only -- cannot weaken above)

OBJECTIVE Mission Objective Rules (select a profile)

Select a mission profile to view its objective rules. Objective rules apply soft boosts/penalties and do not override constraint rules.

USER Add Rule Override

Active User Rules

No user rules applied. System doctrine rules are active.

Impact Preview

Apply a rule change to see which targets change decisions.

Loading rules...

OODA Evaluation Summary

Multi-Cycle Ablation

Click "Load Evaluation Results" to view OODA multi-cycle evaluation across 40 scenarios.

Per-Scenario Results

Edge Activation Frequency

Flag Agreement (DAG vs Flat)

Mission Profile Comparison OBJECTIVE

Compare pipeline results across mission profiles. Constraint rules remain invariant; only objective scoring changes.

Click "Load Profile Comparison" to compare mission profiles.

Objective Parameter Calibration OBJECTIVE learnable

Calibrated objective parameters vs hand-set values. Constraint rules are never modified by calibration.

Click "Load Calibration Results" to view parameter calibration.

ML Model Comparison CLASSIFIER

CNN vs LR through identical doctrine DAG across 40 scenarios. Demonstrates ML-agnostic composition: safety invariants hold regardless of classifier quality.

Click "Load ML Comparison" to compare CNN vs LR pipeline results.

AR-ML Integration

How AR doctrine rules shape CNN training and inference within the OODA loop.

1. AR Cost-Weighted Training

Doctrine rules define which misclassifications matter. BDA feedback from prior cycles adjusts cost weights.

2. AR Adaptive Thresholds

Doctrine context adjusts confidence requirements. Re-observed targets may get adjusted thresholds.

3. BDA Feedback Loop

Engagement outcomes feed back into the next OODA cycle. Relocated targets trigger re-observation and re-classification.

Doctrine Cost Matrix

Misclassification costs derived from engagement rules. Higher cost = CNN pays more attention to avoiding this confusion.

Adaptive Confidence Thresholds

Each target's required confidence level, computed from doctrine context.

Hard Examples for Retraining

Targets where AR overrode CNN's confident prediction -- active learning candidates.

Training Comparison: Standard vs AR-Weighted

AR-Governed Training Profiles OBJECTIVE

Training results per mission profile. The cost matrix is derived from AR doctrine rules, not hand-coded. Constraint violations must remain at zero across all profiles.

Click "Load Training Profiles" to view AR-governed training results.

OODA Cycle Timeline

Cycle-by-cycle breakdown of the closed-loop OODA engagement process. Each cycle includes observe, orient, decide, act, and assess phases with BDA feedback.

Run the OODA loop first to see cycle-by-cycle timeline.

ESTV Reduction Curve

Estimated Surviving Target Value (ESTV) plotted across OODA cycles. Shows how iterative engagement with BDA feedback reduces residual threat.

Run the OODA loop first to see ESTV curve.

ESTV Summary

No data yet.

Per-Cycle Breakdown

No data yet.

BDA Tactical Overlay

Tactical map with targets colored by BDA status. Use the cycle slider to step through engagement outcomes.

BDA Map

● Destroyed ● Damaged ● Relocated ● Missed/Surviving ● Unengaged

BDA Status Summary

Run the OODA loop to see BDA status.

Platform Status

No data yet.

Loop Control Rule Trace

Per-cycle evaluation of loop control rules (L1-L7) that determine whether to CONTINUE, TERMINATE, or RE-OBSERVE. The rule that determined the final decision is highlighted.

Run the OODA loop first to see loop control rule trace.