Add complete BDR Agent Factory structure with docs, UI, and capability registry

.vscode/settings.json

@@ -0,0 +1 @@
+{}

README.md

@@ -3,59 +3,58 @@ title: BDR Agent Factory
 emoji: 🏭
 colorFrom: blue
 colorTo: purple
-sdk: static
+sdk: gradio
+sdk_version: 4.44.0
+app_file: app.py
 pinned: false
-license: mit
-short_description: Enterprise Decision Intelligence Architecture
+license: apache-2.0
+short_description: Enterprise-grade capability registry, governance hub, and system map for Bader AI
 ---
 
-# 🏭 BDR Agent Factory v1
+# 🏭 BDR Agent Factory — Capability Registry & Governance
 
-**Enterprise Decision Intelligence Architecture for Insurance & Regulated Markets**
+**Enterprise-grade capability registry, governance hub, and system map for Bader AI**  
+**GCC Insurance Decision Intelligence Platform**
 
-Architectural foundation for all BDR AI Organization products.
+---
 
-## 🎯 Products
+## What This Space Is
 
-- [ClaimsGPT](https://huggingface.co/spaces/bdr-ai-org/ClaimsGPT) - 62% workload reduction
-- [FraudSimulator-AI](https://huggingface.co/spaces/bdr-ai-org/FraudSimulator-AI) - +15% precision
-- [AutoRiskScoreEngine](https://huggingface.co/spaces/bdr-ai-org/AutoRiskScoreEngine) - IFRS-ready
-- [InsuranceKnowledgeAgent](https://huggingface.co/spaces/bdr-ai-org/InsuranceKnowledgeAgent) - RAG-powered
+BDR Agent Factory is the **official capability registry, governance hub, and system map** for the Bader AI platform.
 
-## 🏗️ Design Principles
+**This is not a demo. This is not a model. This is not a chatbot.**
+
+This is a **reference documentation platform** and **capability browser** that maps AI capabilities to real insurance decision systems, models, datasets, and governance standards.
 
-1. Decision-first, not model-first
-2. PoC 
-cat > README.md << 'ENDREADME'
----
-title: BDR Agent Factory
-emoji: 🏭
-colorFrom: blue
-colorTo: purple
-sdk: static
-pinned: false
-license: mit
-short_description: Enterprise Decision Intelligence Architecture
 ---
 
-# 🏭 BDR Agent Factory v1
+## Capability Dictionary (A–N)
+
+Browse **33+ AI capabilities** organized into **14 categories**:
 
-**Enterprise Decision Intelligence Architecture**
+- **A — NLP**: Text Classification, NER, Summarization, Question Answering
+- **B — Computer Vision**: Image Classification, Object Detection, OCR
+- **G — Analytics**: Anomaly Detection, Risk Scoring, Financial Analysis
+- **I — Evaluation**: Model Evaluation, Explainability (XAI), Bias Detection
+- **N — Governance**: Audit Logging, PII Handling, Drift Monitoring
 
-Architectural foundation for BDR AI Organization products.
+See **[Full Capability Dictionary](docs/01_CAPABILITY_DICTIONARY.md)**
 
-## Products
+---
+
+## Linked Systems
 
-- ClaimsGPT - 62% workload reduction
-- FraudSimulator-AI - +15% precision  
-- AutoRiskScoreEngine - IFRS-ready
-- InsuranceKnowledgeAgent - RAG-powered
+- **[ClaimsGPT](https://huggingface.co/spaces/bdr-ai-org/ClaimsGPT)**: Claim decision intelligence
+- **[FraudSimulator-AI](https://huggingface.co/spaces/bdr-ai-org/FraudSimulator-AI)**: Fraud risk detection
+- **[AutoRiskScoreEngine](https://huggingface.co/spaces/bdr-ai-org/AutoRiskScoreEngine)**: Underwriting risk
+- **[InsuranceKnowledgeAgent](https://huggingface.co/spaces/bdr-ai-org/InsuranceKnowledgeAgent)**: Policy knowledge
 
-## Design Principles
+---
 
-1. Decision-first, not model-first
-2. PoC to MVP to PRD lifecycle
-3. Human-in-the-loop by default
-4. Auditability and governance
+## Documentation
 
-View full documentation at index.html
+- **[Overview](docs/00_OVERVIEW.md)**: What is BDR Agent Factory?
+- **[Capability Dictionary](docs/01_CAPABILITY_DICTIONARY.md)**: Full catalog (A–N)
+- **[Capability Map](docs/02_CAPABILITY_MAP.md)**: System mappings
+- **[Governance](docs/03_GOVERNANCE.md)**: Compliance standards
+- **[Extension Guide](docs/04_EXTENSION_GUIDE.md)**: Add capabilities

app.py

@@ -0,0 +1,9 @@
+import gradio as gr
+from ui.capability_browser import create_capability_browser
+
+# Create the main application
+demo = create_capability_browser()
+
+# Launch the application
+if __name__ == "__main__":
+    demo.launch()

data/capability_registry.json

@@ -0,0 +1,309 @@
+{
+  "capabilities": [
+    {
+      "capability_name": "Text Classification",
+      "category": "A - NLP",
+      "used_in_spaces": ["ClaimsGPT"],
+      "models": ["claims-decision-agent"],
+      "datasets": ["claims-synthetic-dataset"],
+      "insurance_decisions": ["Classify claim type and urgency"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Named Entity Recognition",
+      "category": "A - NLP",
+      "used_in_spaces": ["ClaimsGPT", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Extract claimant info, policy details"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Text Summarization",
+      "category": "A - NLP",
+      "used_in_spaces": ["ClaimsGPT", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Summarize claim reports, policy clauses"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Question Answering",
+      "category": "A - NLP",
+      "used_in_spaces": ["InsuranceKnowledgeAgent"],
+      "models": ["insurance-knowledge-agent"],
+      "datasets": ["insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Answer policy coverage questions"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Sentiment Analysis",
+      "category": "A - NLP",
+      "used_in_spaces": ["ClaimsGPT"],
+      "models": ["claims-decision-agent"],
+      "datasets": ["claims-synthetic-dataset"],
+      "insurance_decisions": ["Analyze customer feedback tone"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "OCR",
+      "category": "B - Computer Vision",
+      "used_in_spaces": ["ClaimsGPT"],
+      "models": ["claims-decision-agent"],
+      "datasets": ["claims-synthetic-dataset"],
+      "insurance_decisions": ["Extract text from claim documents"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Image Classification",
+      "category": "B - Computer Vision",
+      "used_in_spaces": ["ClaimsGPT"],
+      "models": ["claims-decision-agent"],
+      "datasets": ["claims-synthetic-dataset"],
+      "insurance_decisions": ["Classify damage severity from photos"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Object Detection",
+      "category": "B - Computer Vision",
+      "used_in_spaces": ["ClaimsGPT"],
+      "models": ["claims-decision-agent"],
+      "datasets": ["claims-synthetic-dataset"],
+      "insurance_decisions": ["Detect vehicles/damage in claim photos"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Document Understanding",
+      "category": "D - Multimodal AI",
+      "used_in_spaces": ["ClaimsGPT", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Parse claim forms and policy documents"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Text Generation",
+      "category": "E - Generative AI",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Generate decision rationales and explanations"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Semantic Search",
+      "category": "F - Retrieval & Search",
+      "used_in_spaces": ["InsuranceKnowledgeAgent"],
+      "models": ["insurance-knowledge-agent"],
+      "datasets": ["insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Find relevant policy clauses"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "RAG",
+      "category": "F - Retrieval & Search",
+      "used_in_spaces": ["InsuranceKnowledgeAgent"],
+      "models": ["insurance-knowledge-agent"],
+      "datasets": ["insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Retrieve and generate policy answers"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Document Retrieval",
+      "category": "F - Retrieval & Search",
+      "used_in_spaces": ["ClaimsGPT", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Find similar past claims or policy precedents"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Anomaly Detection",
+      "category": "G - Data, Analytics & Visualization",
+      "used_in_spaces": ["FraudSimulator-AI"],
+      "models": ["fraud-risk-agent"],
+      "datasets": ["fraud-simulator-dataset"],
+      "insurance_decisions": ["Detect unusual claim patterns"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Risk Scoring",
+      "category": "G - Data, Analytics & Visualization",
+      "used_in_spaces": ["FraudSimulator-AI", "AutoRiskScoreEngine"],
+      "models": ["fraud-risk-agent", "underwriting-risk-agent"],
+      "datasets": ["fraud-simulator-dataset", "underwriting-risk-dataset"],
+      "insurance_decisions": ["Score fraud risk and underwriting risk"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Financial Analysis",
+      "category": "G - Data, Analytics & Visualization",
+      "used_in_spaces": ["FraudSimulator-AI", "AutoRiskScoreEngine"],
+      "models": ["fraud-risk-agent", "underwriting-risk-agent"],
+      "datasets": ["fraud-simulator-dataset", "underwriting-risk-dataset"],
+      "insurance_decisions": ["Analyze claim amounts and policy financials"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Tabular Classification",
+      "category": "H - Tabular & Structured Data",
+      "used_in_spaces": ["AutoRiskScoreEngine"],
+      "models": ["underwriting-risk-agent"],
+      "datasets": ["underwriting-risk-dataset"],
+      "insurance_decisions": ["Classify policies by risk tier"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Tabular Regression",
+      "category": "H - Tabular & Structured Data",
+      "used_in_spaces": ["AutoRiskScoreEngine"],
+      "models": ["underwriting-risk-agent"],
+      "datasets": ["underwriting-risk-dataset"],
+      "insurance_decisions": ["Predict policy premium or claim amount"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Feature Engineering",
+      "category": "H - Tabular & Structured Data",
+      "used_in_spaces": ["FraudSimulator-AI", "AutoRiskScoreEngine"],
+      "models": ["fraud-risk-agent", "underwriting-risk-agent"],
+      "datasets": ["fraud-simulator-dataset", "underwriting-risk-dataset"],
+      "insurance_decisions": ["Create risk indicators from raw data"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Model Evaluation",
+      "category": "I - Models, Benchmarks & Evaluation",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Validate model performance before deployment"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Explainability (XAI)",
+      "category": "I - Models, Benchmarks & Evaluation",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Explain all decisions with evidence"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Bias Detection",
+      "category": "I - Models, Benchmarks & Evaluation",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Ensure fairness in all decisions"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Model Monitoring",
+      "category": "I - Models, Benchmarks & Evaluation",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Detect model drift over time"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Decision Support Systems",
+      "category": "J - Recommendation & Decision Systems",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset"],
+      "insurance_decisions": ["Recommend approve/reject/escalate/investigate"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Scenario Simulation",
+      "category": "J - Recommendation & Decision Systems",
+      "used_in_spaces": ["FraudSimulator-AI"],
+      "models": ["fraud-risk-agent"],
+      "datasets": ["fraud-simulator-dataset"],
+      "insurance_decisions": ["Simulate fraud scenarios for testing"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Knowledge Graph Construction",
+      "category": "L - Knowledge Graphs & Reasoning",
+      "used_in_spaces": ["FraudSimulator-AI"],
+      "models": ["fraud-risk-agent"],
+      "datasets": ["fraud-simulator-dataset"],
+      "insurance_decisions": ["Map claim relationships to detect fraud rings"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Logical Reasoning",
+      "category": "L - Knowledge Graphs & Reasoning",
+      "used_in_spaces": ["InsuranceKnowledgeAgent"],
+      "models": ["insurance-knowledge-agent"],
+      "datasets": ["insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Apply policy rules to determine coverage"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Workflow Orchestration",
+      "category": "M - Automation & Workflow",
+      "used_in_spaces": ["ClaimsGPT"],
+      "models": ["claims-decision-agent"],
+      "datasets": ["claims-synthetic-dataset"],
+      "insurance_decisions": ["Orchestrate multi-step claim processing"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Audit Logging",
+      "category": "N - Security, Governance & Compliance",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Log all decisions for regulatory audit"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Data Privacy (PII Handling)",
+      "category": "N - Security, Governance & Compliance",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Protect sensitive claimant information"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Drift Monitoring",
+      "category": "N - Security, Governance & Compliance",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Monitor for data/model drift"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Bias Monitoring",
+      "category": "N - Security, Governance & Compliance",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Continuously check for unfair biases"],
+      "governance_required": true
+    },
+    {
+      "capability_name": "Regulatory Compliance",
+      "category": "N - Security, Governance & Compliance",
+      "used_in_spaces": ["ClaimsGPT", "FraudSimulator-AI", "AutoRiskScoreEngine", "InsuranceKnowledgeAgent"],
+      "models": ["claims-decision-agent", "fraud-risk-agent", "underwriting-risk-agent", "insurance-knowledge-agent"],
+      "datasets": ["claims-synthetic-dataset", "fraud-simulator-dataset", "underwriting-risk-dataset", "insurance-policy-docs-dataset"],
+      "insurance_decisions": ["Ensure IFRS, AML, GCC compliance"],
+      "governance_required": true
+    }
+  ],
+  "metadata": {
+    "version": "1.0.0",
+    "last_updated": "2026-01-01",
+    "total_capabilities": 33,
+    "total_systems": 4,
+    "total_models": 4,
+    "total_datasets": 4
+  }
+}

docs/00_OVERVIEW.md

@@ -0,0 +1,129 @@
+# BDR Agent Factory — Overview
+
+## What is Bader AI?
+
+Bader AI is an **Enterprise Decision Intelligence Platform** designed specifically for the insurance industry in the GCC region. It transforms raw data into governed, auditable, and accountable decisions across claims processing, fraud detection, underwriting risk assessment, and policy knowledge operations.
+
+Bader AI is not a collection of AI models. It is a **decision-first architecture** where every AI capability exists to serve a specific, measurable insurance decision outcome.
+
+## What is BDR Agent Factory?
+
+BDR Agent Factory is the **official capability registry, governance hub, and system map** for the Bader AI platform.
+
+This Space serves as:
+
+- **Capability Dictionary**: A comprehensive catalog of all AI capabilities (A–N categories) used across the platform
+- **Capability-to-System Map**: Explicit mappings showing which capabilities power which insurance decision systems
+- **Governance Standards**: Unified policies for auditability, explainability, and regulatory compliance
+- **Extension Guide**: Rules and procedures for adding new capabilities without duplication
+
+**This is not a demo. This is not a model. This is not a chatbot.**
+
+This is a **reference documentation platform** and **capability browser** that maps AI capabilities to real insurance decision systems, models, datasets, and governance standards.
+
+## AI Capability vs Insurance System
+
+### AI Capability
+A **capability** is a technical function or method that enables a specific type of AI processing.
+
+Examples:
+- Anomaly Detection
+- Named Entity Recognition
+- Risk Scoring
+- Document Classification
+
+Capabilities are **reusable building blocks**.
+
+### Insurance System
+A **system** is a complete decision intelligence application that answers a specific executive question.
+
+Examples:
+- **ClaimsGPT**: "Should this claim be approved, rejected, or escalated?"
+- **FraudSimulator-AI**: "Should this claim be investigated or allowed?"
+- **AutoRiskScoreEngine**: "What risk band does this policy belong to?"
+- **InsuranceKnowledgeAgent**: "Does this clause apply to this scenario?"
+
+Systems **compose multiple capabilities** to deliver a decision.
+
+## Why This Registry Exists
+
+### Problem
+Without a central registry:
+- Capabilities get duplicated across projects
+- Systems become black boxes
+- Governance is inconsistent
+- Regulators cannot audit the platform
+- CTOs cannot understand the architecture
+
+### Solution
+BDR Agent Factory provides:
+- **Single source of truth** for all capabilities
+- **Explicit mappings** from capabilities → systems → models → datasets → decisions
+- **Governance standards** that apply uniformly across all systems
+- **Transparency** for regulators, executives, and technical teams
+
+## How to Read and Use This Registry
+
+### For Regulators
+1. Review the **Capability Dictionary** to understand what AI methods are used
+2. Check the **Capability Map** to see which systems use which capabilities
+3. Examine the **Governance Standards** to verify compliance controls
+4. Audit decision logs and explainability outputs from linked systems
+
+### For CTOs and Technical Leaders
+1. Use the **Capability Dictionary** as a reference for available AI functions
+2. Consult the **Capability Map** before building new systems to avoid duplication
+3. Follow the **Extension Guide** when adding new capabilities
+4. Ensure all systems comply with **Governance Standards**
+
+### For Insurers and Business Users
+1. Understand the **difference between capabilities and systems**
+2. Review the **Capability Map** to see how decisions are made
+3. Trust that all systems follow **unified governance standards**
+4. Use linked Spaces (ClaimsGPT, FraudSimulator-AI, etc.) for actual decision-making
+
+## Core Principles
+
+### 1. Decision-First, Not Model-First
+Every capability exists only if it serves an insurance decision. No capability is created in isolation.
+
+### 2. No Duplicated Capabilities
+If a capability already exists in the registry, it must be reused. No standalone capability projects are allowed.
+
+### 3. Single Source of Truth
+This registry is the authoritative reference for:
+- Capabilities
+- Systems
+- Models
+- Datasets
+- Governance standards
+
+### 4. Hugging Face Native
+All systems, models, and datasets are hosted on Hugging Face for transparency and reproducibility.
+
+### 5. GCC Insurance Context
+All capabilities and systems are designed for:
+- IFRS compliance
+- AML (Anti-Money Laundering) requirements
+- Takaful-ready operations
+- GCC regulatory standards
+
+## Linked Systems
+
+The following Hugging Face Spaces are governed by this registry:
+
+- **[ClaimsGPT](https://huggingface.co/spaces/bdr-ai-org/ClaimsGPT)**: AI-powered claim decision intelligence with document analysis
+- **[FraudSimulator-AI](https://huggingface.co/spaces/bdr-ai-org/FraudSimulator-AI)**: Fraud risk and anomaly decision intelligence
+- **[AutoRiskScoreEngine](https://huggingface.co/spaces/bdr-ai-org/AutoRiskScoreEngine)**: IFRS-ready underwriting risk assessment
+- **[InsuranceKnowledgeAgent](https://huggingface.co/spaces/bdr-ai-org/InsuranceKnowledgeAgent)**: RAG-powered policy knowledge and clause resolution
+
+## Navigation
+
+- **[Capability Dictionary](01_CAPABILITY_DICTIONARY.md)**: Full catalog of AI capabilities (A–N)
+- **[Capability Map](02_CAPABILITY_MAP.md)**: Mappings from capabilities to systems, models, datasets, and decisions
+- **[Governance Standards](03_GOVERNANCE.md)**: Unified governance policies
+- **[Extension Guide](04_EXTENSION_GUIDE.md)**: How to add new capabilities
+
+---
+
+**BDR Agent Factory** — The authoritative capability registry for Bader AI, the GCC Insurance Decision Intelligence Platform.

docs/01_CAPABILITY_DICTIONARY.md

@@ -0,0 +1,290 @@
+# AI Capability Dictionary (A–N)
+
+This document provides the **complete catalog of AI capabilities** used across the Bader AI platform. Each capability is categorized, defined, and contextualized for insurance decision intelligence.
+
+---
+
+## Category A — Natural Language Processing (NLP)
+
+### Text Classification
+**Definition**: Categorizing text into predefined classes or labels.  
+**Insurance Relevance**: Classify claim descriptions, policy documents, customer inquiries by type, urgency, or risk level.
+
+### Named Entity Recognition (NER)
+**Definition**: Identifying and extracting entities (names, dates, locations, amounts) from unstructured text.  
+**Insurance Relevance**: Extract claimant names, incident dates, locations, and monetary amounts from claim forms and reports.
+
+### Sentiment Analysis
+**Definition**: Determining the emotional tone or sentiment expressed in text.  
+**Insurance Relevance**: Analyze customer feedback, complaint letters, or social media mentions to gauge satisfaction and identify escalation risks.
+
+### Text Summarization
+**Definition**: Generating concise summaries of longer documents.  
+**Insurance Relevance**: Summarize lengthy claim reports, policy documents, or investigation notes for quick executive review.
+
+### Question Answering
+**Definition**: Providing direct answers to questions based on a given context or knowledge base.  
+**Insurance Relevance**: Answer policyholder questions about coverage, exclusions, or claim status using policy documents as context.
+
+### Language Translation
+**Definition**: Converting text from one language to another.  
+**Insurance Relevance**: Translate claim documents, policy terms, or customer communications across Arabic, English, and other GCC languages.
+
+---
+
+## Category B — Computer Vision
+
+### Image Classification
+**Definition**: Categorizing images into predefined classes.  
+**Insurance Relevance**: Classify damage photos (vehicle, property) by severity or type (e.g., minor dent, total loss).
+
+### Object Detection
+**Definition**: Identifying and locating objects within images.  
+**Insurance Relevance**: Detect vehicles, property damage, or specific items in claim photos to verify incident details.
+
+### Image Segmentation
+**Definition**: Partitioning an image into multiple segments or regions.  
+**Insurance Relevance**: Segment damaged areas in property or vehicle images to assess repair scope.
+
+### Optical Character Recognition (OCR)
+**Definition**: Extracting text from images or scanned documents.  
+**Insurance Relevance**: Digitize handwritten claim forms, invoices, medical reports, or ID documents for automated processing.
+
+### Facial Recognition
+**Definition**: Identifying or verifying individuals based on facial features.  
+**Insurance Relevance**: Verify claimant identity during video claims or detect duplicate claims from the same individual.
+
+---
+
+## Category C — Audio & Speech
+
+### Speech Recognition
+**Definition**: Converting spoken language into text.  
+**Insurance Relevance**: Transcribe customer service calls, claim interviews, or voice-based claim submissions.
+
+### Text-to-Speech (TTS)
+**Definition**: Converting text into spoken audio.  
+**Insurance Relevance**: Provide voice-based policy summaries, claim status updates, or accessibility features for visually impaired users.
+
+### Speaker Identification
+**Definition**: Identifying who is speaking in an audio recording.  
+**Insurance Relevance**: Verify caller identity in phone-based claims or detect fraudulent impersonation attempts.
+
+### Audio Classification
+**Definition**: Categorizing audio clips by type or content.  
+**Insurance Relevance**: Classify call center recordings by topic (claim inquiry, complaint, policy question) for routing and analysis.
+
+---
+
+## Category D — Multimodal AI
+
+### Vision-Language Models
+**Definition**: Models that process both images and text to understand and generate content.  
+**Insurance Relevance**: Analyze claim photos alongside written descriptions to verify consistency and detect discrepancies.
+
+### Document Understanding
+**Definition**: Extracting structured information from complex documents (forms, invoices, contracts).  
+**Insurance Relevance**: Parse insurance claim forms, medical bills, repair invoices, and policy contracts for automated data entry.
+
+### Visual Question Answering
+**Definition**: Answering questions about the content of an image.  
+**Insurance Relevance**: Answer questions like "Is the damage visible in this photo?" or "What type of vehicle is shown?"
+
+---
+
+## Category E — Generative AI
+
+### Text Generation
+**Definition**: Creating human-like text based on prompts or context.  
+**Insurance Relevance**: Generate claim summaries, policy explanations, customer communications, or investigation reports.
+
+### Image Generation
+**Definition**: Creating images from text descriptions or other inputs.  
+**Insurance Relevance**: Generate visual aids for policy explanations or training materials (limited use in production decisions).
+
+### Code Generation
+**Definition**: Automatically generating code from natural language descriptions.  
+**Insurance Relevance**: Automate report generation scripts, data transformation pipelines, or decision logic implementations.
+
+---
+
+## Category F — Retrieval & Search
+
+### Semantic Search
+**Definition**: Finding information based on meaning rather than exact keyword matches.  
+**Insurance Relevance**: Search policy documents, claim histories, or knowledge bases using natural language queries.
+
+### Retrieval-Augmented Generation (RAG)
+**Definition**: Combining retrieval of relevant documents with generative AI to produce informed responses.  
+**Insurance Relevance**: Answer policy questions by retrieving relevant clauses and generating contextual explanations.
+
+### Document Retrieval
+**Definition**: Finding relevant documents from a large corpus based on a query.  
+**Insurance Relevance**: Retrieve similar past claims, precedent cases, or relevant policy sections for decision support.
+
+---
+
+## Category G — Data, Analytics & Visualization
+
+### Anomaly Detection
+**Definition**: Identifying unusual patterns or outliers in data.  
+**Insurance Relevance**: Detect fraudulent claims, unusual claim patterns, or data entry errors.
+
+### Time Series Analysis
+**Definition**: Analyzing data points collected over time to identify trends or patterns.  
+**Insurance Relevance**: Forecast claim volumes, detect seasonal fraud patterns, or predict policy renewals.
+
+### Risk Scoring
+**Definition**: Assigning numerical risk scores based on multiple factors.  
+**Insurance Relevance**: Score claims, policies, or customers by fraud risk, underwriting risk, or churn probability.
+
+### Financial Analysis
+**Definition**: Analyzing financial data to derive insights or forecasts.  
+**Insurance Relevance**: Assess claim reserve adequacy, policy profitability, or loss ratios.
+
+### Data Visualization
+**Definition**: Creating visual representations of data (charts, graphs, dashboards).  
+**Insurance Relevance**: Visualize claim trends, fraud patterns, risk distributions, or portfolio performance.
+
+---
+
+## Category H — Tabular & Structured Data
+
+### Tabular Classification
+**Definition**: Classifying rows in structured datasets.  
+**Insurance Relevance**: Classify policies by risk tier, claims by approval likelihood, or customers by segment.
+
+### Tabular Regression
+**Definition**: Predicting continuous values from structured data.  
+**Insurance Relevance**: Predict claim amounts, policy premiums, or customer lifetime value.
+
+### Feature Engineering
+**Definition**: Creating new features from raw data to improve model performance.  
+**Insurance Relevance**: Derive features like claim frequency, average claim size, or time since last claim for risk models.
+
+---
+
+## Category I — Models, Benchmarks & Evaluation
+
+### Model Evaluation
+**Definition**: Assessing model performance using metrics (accuracy, precision, recall, F1, AUC).  
+**Insurance Relevance**: Validate fraud detection models, claim approval models, or risk scoring models before deployment.
+
+### Explainability (XAI)
+**Definition**: Providing human-understandable explanations for model predictions.  
+**Insurance Relevance**: Explain why a claim was flagged for fraud, why a policy was rated high-risk, or why a decision was made.
+
+### Bias Detection
+**Definition**: Identifying unfair biases in model predictions.  
+**Insurance Relevance**: Ensure claim decisions, underwriting, and fraud detection do not discriminate based on protected attributes.
+
+### Model Monitoring
+**Definition**: Tracking model performance over time to detect drift or degradation.  
+**Insurance Relevance**: Monitor fraud detection accuracy, claim approval rates, or risk score distributions for drift.
+
+---
+
+## Category J — Recommendation & Decision Systems
+
+### Recommendation Systems
+**Definition**: Suggesting items, actions, or content based on user preferences or context.  
+**Insurance Relevance**: Recommend policy add-ons, coverage adjustments, or next-best actions for claims handlers.
+
+### Decision Support Systems
+**Definition**: Providing data-driven recommendations to assist human decision-making.  
+**Insurance Relevance**: Recommend claim approval/rejection, investigation priority, or settlement amounts with supporting evidence.
+
+### Scenario Simulation
+**Definition**: Modeling hypothetical scenarios to predict outcomes.  
+**Insurance Relevance**: Simulate fraud scenarios, catastrophe impacts, or policy portfolio changes to assess risk.
+
+---
+
+## Category K — Reinforcement Learning
+
+### Policy Optimization
+**Definition**: Learning optimal strategies through trial and error.  
+**Insurance Relevance**: Optimize claim routing, fraud investigation prioritization, or customer engagement strategies.
+
+### Multi-Armed Bandits
+**Definition**: Balancing exploration and exploitation to maximize rewards.  
+**Insurance Relevance**: Optimize A/B testing for claim workflows, pricing strategies, or customer communications.
+
+---
+
+## Category L — Knowledge Graphs & Reasoning
+
+### Knowledge Graph Construction
+**Definition**: Building structured representations of entities and relationships.  
+**Insurance Relevance**: Map relationships between claimants, policies, incidents, and providers to detect fraud rings.
+
+### Logical Reasoning
+**Definition**: Applying rules and logic to derive conclusions.  
+**Insurance Relevance**: Apply policy rules, coverage conditions, and exclusions to determine claim eligibility.
+
+### Ontology Alignment
+**Definition**: Mapping concepts across different knowledge systems.  
+**Insurance Relevance**: Align internal policy terms with regulatory definitions or industry standards.
+
+---
+
+## Category M — Automation & Workflow
+
+### Robotic Process Automation (RPA)
+**Definition**: Automating repetitive, rule-based tasks.  
+**Insurance Relevance**: Automate data entry, document routing, or status updates in claim processing.
+
+### Workflow Orchestration
+**Definition**: Coordinating multi-step processes across systems and agents.  
+**Insurance Relevance**: Orchestrate claim intake, validation, investigation, approval, and payment workflows.
+
+### Task Scheduling
+**Definition**: Automatically scheduling tasks based on priorities and dependencies.  
+**Insurance Relevance**: Schedule claim reviews, fraud investigations, or policy renewals based on urgency and capacity.
+
+---
+
+## Category N — Security, Governance & Compliance
+
+### Audit Logging
+**Definition**: Recording all system actions and decisions for accountability.  
+**Insurance Relevance**: Log every claim decision, model prediction, and user action for regulatory audits.
+
+### Access Control
+**Definition**: Managing who can access what data or perform what actions.  
+**Insurance Relevance**: Restrict access to sensitive claim data, PII, or financial information based on roles.
+
+### Data Privacy (PII Handling)
+**Definition**: Protecting personally identifiable information.  
+**Insurance Relevance**: Anonymize, encrypt, or redact PII in claim documents, customer records, and analytics.
+
+### Drift Monitoring
+**Definition**: Detecting changes in data distributions or model behavior over time.  
+**Insurance Relevance**: Detect shifts in claim patterns, fraud tactics, or customer behavior that may degrade model performance.
+
+### Bias Monitoring
+**Definition**: Continuously checking for unfair biases in model outputs.  
+**Insurance Relevance**: Ensure ongoing fairness in claim approvals, underwriting, and fraud detection.
+
+### Regulatory Compliance
+**Definition**: Ensuring systems meet legal and regulatory requirements.  
+**Insurance Relevance**: Comply with IFRS, AML, GDPR, and GCC insurance regulations in all decision systems.
+
+---
+
+## Summary
+
+This dictionary contains **60+ AI capabilities** across **14 categories (A–N)**. Each capability is:
+- **Defined** clearly
+- **Contextualized** for insurance use cases
+- **Reusable** across multiple systems
+- **Governed** by unified standards
+
+**Next Steps**:
+- See **[Capability Map](02_CAPABILITY_MAP.md)** to understand which capabilities power which systems
+- Review **[Governance Standards](03_GOVERNANCE.md)** to understand how capabilities are governed
+- Consult **[Extension Guide](04_EXTENSION_GUIDE.md)** before adding new capabilities
+
+---
+
+**BDR Agent Factory** — The authoritative capability registry for Bader AI.

docs/02_CAPABILITY_MAP.md

@@ -0,0 +1,237 @@
+# Capability Map — Systems, Models, Datasets, Decisions
+
+This document provides **explicit mappings** from AI capabilities to the systems, models, datasets, and insurance decisions they support.
+
+---
+
+## Mapping Table
+
+| Capability | Category | Used In Space | Model | Dataset | Insurance Decision |
+|-----------|----------|---------------|-------|---------|-------------------|
+| **Text Classification** | A - NLP | ClaimsGPT | claims-decision-agent | claims-synthetic-dataset | Classify claim type and urgency |
+| **Named Entity Recognition** | A - NLP | ClaimsGPT, InsuranceKnowledgeAgent | claims-decision-agent, insurance-knowledge-agent | claims-synthetic-dataset, insurance-policy-docs-dataset | Extract claimant info, policy details |
+| **Text Summarization** | A - NLP | ClaimsGPT, InsuranceKnowledgeAgent | claims-decision-agent, insurance-knowledge-agent | claims-synthetic-dataset, insurance-policy-docs-dataset | Summarize claim reports, policy clauses |
+| **Question Answering** | A - NLP | InsuranceKnowledgeAgent | insurance-knowledge-agent | insurance-policy-docs-dataset | Answer policy coverage questions |
+| **OCR** | B - Vision | ClaimsGPT | claims-decision-agent | claims-synthetic-dataset | Extract text from claim documents |
+| **Image Classification** | B - Vision | ClaimsGPT | claims-decision-agent | claims-synthetic-dataset | Classify damage severity from photos |
+| **Object Detection** | B - Vision | ClaimsGPT | claims-decision-agent | claims-synthetic-dataset | Detect vehicles/damage in claim photos |
+| **Document Understanding** | D - Multimodal | ClaimsGPT, InsuranceKnowledgeAgent | claims-decision-agent, insurance-knowledge-agent | claims-synthetic-dataset, insurance-policy-docs-dataset | Parse claim forms and policy documents |
+| **Text Generation** | E - Generative | ClaimsGPT, FraudSimulator-AI, InsuranceKnowledgeAgent | claims-decision-agent, fraud-risk-agent, insurance-knowledge-agent | claims-synthetic-dataset, fraud-simulator-dataset, insurance-policy-docs-dataset | Generate decision rationales and explanations |
+| **Semantic Search** | F - Retrieval | InsuranceKnowledgeAgent | insurance-knowledge-agent | insurance-policy-docs-dataset | Find relevant policy clauses |
+| **RAG** | F - Retrieval | InsuranceKnowledgeAgent | insurance-knowledge-agent | insurance-policy-docs-dataset | Retrieve and generate policy answers |
+| **Document Retrieval** | F - Retrieval | ClaimsGPT, InsuranceKnowledgeAgent | claims-decision-agent, insurance-knowledge-agent | claims-synthetic-dataset, insurance-policy-docs-dataset | Find similar past claims or policy precedents |
+| **Anomaly Detection** | G - Analytics | FraudSimulator-AI | fraud-risk-agent | fraud-simulator-dataset | Detect unusual claim patterns |
+| **Risk Scoring** | G - Analytics | FraudSimulator-AI, AutoRiskScoreEngine | fraud-risk-agent, underwriting-risk-agent | fraud-simulator-dataset, underwriting-risk-dataset | Score fraud risk and underwriting risk |
+| **Financial Analysis** | G - Analytics | FraudSimulator-AI, AutoRiskScoreEngine | fraud-risk-agent, underwriting-risk-agent | fraud-simulator-dataset, underwriting-risk-dataset | Analyze claim amounts and policy financials |
+| **Tabular Classification** | H - Structured Data | AutoRiskScoreEngine | underwriting-risk-agent | underwriting-risk-dataset | Classify policies by risk tier |
+| **Tabular Regression** | H - Structured Data | AutoRiskScoreEngine | underwriting-risk-agent | underwriting-risk-dataset | Predict policy premium or claim amount |
+| **Feature Engineering** | H - Structured Data | FraudSimulator-AI, AutoRiskScoreEngine | fraud-risk-agent, underwriting-risk-agent | fraud-simulator-dataset, underwriting-risk-dataset | Create risk indicators from raw data |
+| **Model Evaluation** | I - Evaluation | All Spaces | All Models | All Datasets | Validate model performance before deployment |
+| **Explainability (XAI)** | I - Evaluation | All Spaces | All Models | All Datasets | Explain all decisions with evidence |
+| **Bias Detection** | I - Evaluation | All Spaces | All Models | All Datasets | Ensure fairness in all decisions |
+| **Model Monitoring** | I - Evaluation | All Spaces | All Models | All Datasets | Detect model drift over time |
+| **Decision Support Systems** | J - Recommendations | ClaimsGPT, FraudSimulator-AI, AutoRiskScoreEngine | claims-decision-agent, fraud-risk-agent, underwriting-risk-agent | claims-synthetic-dataset, fraud-simulator-dataset, underwriting-risk-dataset | Recommend approve/reject/escalate/investigate |
+| **Scenario Simulation** | J - Recommendations | FraudSimulator-AI | fraud-risk-agent | fraud-simulator-dataset | Simulate fraud scenarios for testing |
+| **Knowledge Graph Construction** | L - Knowledge | FraudSimulator-AI | fraud-risk-agent | fraud-simulator-dataset | Map claim relationships to detect fraud rings |
+| **Logical Reasoning** | L - Knowledge | InsuranceKnowledgeAgent | insurance-knowledge-agent | insurance-policy-docs-dataset | Apply policy rules to determine coverage |
+| **Workflow Orchestration** | M - Automation | ClaimsGPT | claims-decision-agent | claims-synthetic-dataset | Orchestrate multi-step claim processing |
+| **Audit Logging** | N - Governance | All Spaces | All Models | All Datasets | Log all decisions for regulatory audit |
+| **Data Privacy (PII Handling)** | N - Governance | All Spaces | All Models | All Datasets | Protect sensitive claimant information |
+| **Drift Monitoring** | N - Governance | All Spaces | All Models | All Datasets | Monitor for data/model drift |
+| **Bias Monitoring** | N - Governance | All Spaces | All Models | All Datasets | Continuously check for unfair biases |
+| **Regulatory Compliance** | N - Governance | All Spaces | All Models | All Datasets | Ensure IFRS, AML, GCC compliance |
+
+---
+
+## System-Specific Capability Breakdown
+
+### ClaimsGPT
+**Decision**: Approve | Reject | Escalate
+
+**Capabilities Used**:
+- Text Classification (claim type)
+- Named Entity Recognition (extract claimant info)
+- Text Summarization (summarize reports)
+- OCR (digitize documents)
+- Image Classification (damage severity)
+- Object Detection (verify incident details)
+- Document Understanding (parse claim forms)
+- Text Generation (generate decision rationale)
+- Document Retrieval (find similar claims)
+- Decision Support Systems (recommend action)
+- Workflow Orchestration (multi-step processing)
+- Explainability (explain decision)
+- Audit Logging (record decision)
+
+**Model**: `bdr-ai-org/claims-decision-agent`  
+**Dataset**: `bdr-ai-org/claims-synthetic-dataset`
+
+---
+
+### FraudSimulator-AI
+**Decision**: Investigate | Allow
+
+**Capabilities Used**:
+- Anomaly Detection (detect unusual patterns)
+- Risk Scoring (fraud risk score)
+- Financial Analysis (analyze claim amounts)
+- Feature Engineering (create fraud indicators)
+- Text Generation (generate investigation notes)
+- Scenario Simulation (test fraud scenarios)
+- Knowledge Graph Construction (map claim networks)
+- Decision Support Systems (recommend investigate/allow)
+- Explainability (explain fraud signals)
+- Audit Logging (record fraud decisions)
+
+**Model**: `bdr-ai-org/fraud-risk-agent`  
+**Dataset**: `bdr-ai-org/fraud-simulator-dataset`
+
+---
+
+### AutoRiskScoreEngine
+**Decision**: Price | Segment | Escalate
+
+**Capabilities Used**:
+- Risk Scoring (underwriting risk)
+- Financial Analysis (policy profitability)
+- Tabular Classification (risk tier)
+- Tabular Regression (premium prediction)
+- Feature Engineering (risk indicators)
+- Decision Support Systems (recommend pricing/segment)
+- Explainability (explain risk score)
+- Audit Logging (record underwriting decisions)
+
+**Model**: `bdr-ai-org/underwriting-risk-agent`  
+**Dataset**: `bdr-ai-org/underwriting-risk-dataset`
+
+---
+
+### InsuranceKnowledgeAgent
+**Decision**: Clause Applicability | Coverage Decision
+
+**Capabilities Used**:
+- Named Entity Recognition (extract policy terms)
+- Text Summarization (summarize clauses)
+- Question Answering (answer coverage questions)
+- Semantic Search (find relevant clauses)
+- RAG (retrieve and generate answers)
+- Document Retrieval (find policy precedents)
+- Document Understanding (parse policy documents)
+- Text Generation (generate explanations)
+- Logical Reasoning (apply policy rules)
+- Explainability (explain coverage decisions)
+- Audit Logging (record knowledge queries)
+
+**Model**: `bdr-ai-org/insurance-knowledge-agent`  
+**Dataset**: `bdr-ai-org/insurance-policy-docs-dataset`
+
+---
+
+## Model Contracts
+
+### claims-decision-agent
+```json
+{
+  "decision": "approve | reject | escalate",
+  "confidence": 0.0-1.0,
+  "claim_summary": "string",
+  "extracted_entities": {},
+  "damage_assessment": {},
+  "similar_claims": [],
+  "rationale": "string",
+  "explainability": {
+    "factors": [],
+    "weights": {}
+  }
+}
+```
+
+### fraud-risk-agent
+```json
+{
+  "fraud_score": 0.0-1.0,
+  "risk_band": "low | medium | high",
+  "top_indicators": [],
+  "recommended_action": "investigate | allow",
+  "confidence": 0.0-1.0,
+  "explainability": {
+    "signals": [],
+    "weights": {}
+  }
+}
+```
+
+### underwriting-risk-agent
+```json
+{
+  "risk_score": 0.0-1.0,
+  "risk_tier": "low | medium | high | very_high",
+  "recommended_premium": 0.0,
+  "recommended_action": "price | segment | escalate",
+  "confidence": 0.0-1.0,
+  "explainability": {
+    "risk_factors": [],
+    "weights": {}
+  }
+}
+```
+
+### insurance-knowledge-agent
+```json
+{
+  "answer": "string",
+  "relevant_clauses": [],
+  "coverage_decision": "applicable | not_applicable | unclear",
+  "confidence": 0.0-1.0,
+  "explainability": {
+    "retrieved_sections": [],
+    "reasoning_chain": []
+  }
+}
+```
+
+---
+
+## Dataset Lineage
+
+### claims-synthetic-dataset
+- **Purpose**: Training and validation for claim decision models
+- **Contents**: Synthetic insurance claims with labels (approve/reject/escalate)
+- **Features**: Claim text, claimant info, incident details, damage photos, financial amounts
+- **Governance**: PII anonymized, bias-checked, version-controlled
+
+### fraud-simulator-dataset
+- **Purpose**: Training and stress-testing fraud detection models
+- **Contents**: Normal and fraudulent claim patterns
+- **Features**: Claim frequency, amount deviations, entity relationships, temporal patterns
+- **Governance**: Synthetic data, fraud scenario labels, drift scenarios included
+
+### underwriting-risk-dataset
+- **Purpose**: Training underwriting risk models
+- **Contents**: Policy data with risk labels and premium outcomes
+- **Features**: Policy type, coverage limits, customer demographics, historical claims
+- **Governance**: Anonymized, IFRS-aligned, bias-monitored
+
+### insurance-policy-docs-dataset
+- **Purpose**: Knowledge base for policy question answering
+- **Contents**: Insurance policy documents, clauses, terms, conditions
+- **Features**: Policy text, clause embeddings, metadata (product type, region)
+- **Governance**: Version-controlled, conflict-resolved, Takaful-ready
+
+---
+
+## Governance Alignment
+
+All capabilities, systems, models, and datasets must comply with:
+- **Audit Logging**: Every decision recorded
+- **Explainability**: Every decision explained
+- **Bias Monitoring**: Continuous fairness checks
+- **Drift Monitoring**: Performance tracking over time
+- **PII Handling**: Data privacy and anonymization
+- **Regulatory Compliance**: IFRS, AML, GCC standards
+
+See **[Governance Standards](03_GOVERNANCE.md)** for full details.
+
+---
+
+**BDR Agent Factory** — Explicit mappings for transparent, auditable decision intelligence.

docs/03_GOVERNANCE.md

@@ -0,0 +1,322 @@
+# Governance Standards
+
+This document defines the **unified governance policies** that apply to all systems, models, datasets, and capabilities within the Bader AI platform.
+
+These standards ensure that every decision is **auditable, explainable, fair, and compliant** with regulatory requirements.
+
+---
+
+## Core Governance Principles
+
+### 1. Decision Traceability
+Every decision made by any system must be:
+- **Logged** with a unique decision ID
+- **Timestamped** with UTC time
+- **Attributed** to the specific model version and user (if applicable)
+- **Linked** to the input data snapshot used for the decision
+
+### 2. Explainability by Default
+Every decision must include:
+- **Rationale**: Human-readable explanation of why the decision was made
+- **Evidence**: Key factors, signals, or data points that influenced the decision
+- **Confidence Score**: Numerical confidence level (0.0–1.0)
+- **Alternative Outcomes**: What other decisions were considered and why they were not chosen
+
+### 3. Human-in-the-Loop Readiness
+All systems must support:
+- **Human Override**: Ability for authorized users to override AI decisions
+- **Escalation Paths**: Clear rules for when decisions must be escalated to humans
+- **Feedback Loops**: Mechanisms to capture human corrections and improve models
+
+### 4. Fairness and Bias Mitigation
+All models must be:
+- **Bias-Tested**: Evaluated for unfair biases before deployment
+- **Continuously Monitored**: Ongoing checks for bias in production decisions
+- **Demographically Fair**: No discrimination based on protected attributes (age, gender, nationality, etc.)
+
+### 5. Regulatory Compliance
+All systems must comply with:
+- **IFRS Standards**: International Financial Reporting Standards for insurance
+- **AML Requirements**: Anti-Money Laundering regulations
+- **GCC Regulations**: Local insurance and data protection laws
+- **Takaful Principles**: Islamic insurance compliance where applicable
+
+---
+
+## Governance Standards by Category
+
+### A. Decision Audit Logs
+
+**Requirement**: Every decision must be logged with the following fields:
+
+```json
+{
+  "decision_id": "unique-uuid",
+  "timestamp": "ISO-8601 UTC",
+  "system": "ClaimsGPT | FraudSimulator-AI | AutoRiskScoreEngine | InsuranceKnowledgeAgent",
+  "model_version": "semantic-version",
+  "user_id": "user-identifier (if applicable)",
+  "input_snapshot": {
+    "claim_id": "...",
+    "policy_id": "...",
+    "raw_data": "..."
+  },
+  "decision": "approve | reject | escalate | investigate | allow | price | segment | applicable | not_applicable",
+  "confidence": 0.0-1.0,
+  "rationale": "human-readable explanation",
+  "explainability": {
+    "factors": [],
+    "weights": {},
+    "evidence": []
+  },
+  "human_override": false,
+  "override_reason": null
+}
+```
+
+**Storage**: Audit logs must be stored in an immutable, append-only data store with retention for at least 7 years (regulatory requirement).
+
+**Access**: Audit logs must be accessible to:
+- Regulators (read-only)
+- Internal audit teams (read-only)
+- System administrators (read-only)
+- Authorized compliance officers (read-only)
+
+---
+
+### B. Explainability (XAI)
+
+**Requirement**: Every decision must include explainability outputs.
+
+**Explainability Components**:
+1. **Rationale**: 1-3 sentence summary of why the decision was made
+2. **Top Factors**: 3-5 most influential factors with importance scores
+3. **Evidence**: Specific data points or patterns that supported the decision
+4. **Counterfactuals**: What would need to change for a different decision
+
+**Example (Fraud Detection)**:
+```json
+{
+  "rationale": "This claim was flagged for investigation due to unusually high claim amount relative to policy coverage and multiple claims from the same claimant in a short time period.",
+  "top_factors": [
+    {"factor": "claim_amount_deviation", "importance": 0.45},
+    {"factor": "claim_frequency", "importance": 0.30},
+    {"factor": "entity_linkage", "importance": 0.25}
+  ],
+  "evidence": [
+    "Claim amount: $50,000 vs. average: $5,000",
+    "3 claims in 30 days vs. historical average: 0.2 claims/month",
+    "Claimant linked to 2 other flagged claims"
+  ],
+  "counterfactual": "If claim amount were below $10,000 and claim frequency were normal, decision would be 'allow'."
+}
+```
+
+**Validation**: Explainability outputs must be:
+- **Accurate**: Reflect actual model behavior
+- **Understandable**: Written in plain language, not technical jargon
+- **Actionable**: Provide clear next steps or justification
+
+---
+
+### C. Human-in-the-Loop Override
+
+**Requirement**: All systems must support human override of AI decisions.
+
+**Override Process**:
+1. Authorized user reviews AI decision and explainability
+2. User provides override reason (required)
+3. System logs override with user ID, timestamp, and reason
+4. Override decision takes precedence over AI decision
+5. Override is flagged for model retraining consideration
+
+**Override Logging**:
+```json
+{
+  "decision_id": "original-decision-uuid",
+  "override_timestamp": "ISO-8601 UTC",
+  "override_user_id": "user-identifier",
+  "original_decision": "reject",
+  "override_decision": "approve",
+  "override_reason": "Additional evidence provided by claimant not available to AI model",
+  "override_category": "new_evidence | model_error | policy_exception | other"
+}
+```
+
+**Escalation Rules**:
+- **Low Confidence**: Decisions with confidence < 0.7 must be escalated to human review
+- **High Stakes**: Decisions involving amounts > $X or sensitive cases must be escalated
+- **Conflicting Signals**: Decisions with contradictory evidence must be escalated
+- **Regulatory Flags**: Decisions involving AML or compliance flags must be escalated
+
+---
+
+### D. Bias and Drift Monitoring
+
+**Bias Monitoring**:
+- **Pre-Deployment**: Models must pass bias tests before production deployment
+- **Continuous Monitoring**: Production decisions monitored for demographic disparities
+- **Metrics**: Track approval rates, fraud flags, risk scores across protected groups
+- **Thresholds**: Alert if disparities exceed acceptable thresholds (e.g., >10% difference)
+
+**Bias Testing Checklist**:
+- [ ] Test for gender bias
+- [ ] Test for age bias
+- [ ] Test for nationality bias
+- [ ] Test for geographic bias
+- [ ] Test for socioeconomic bias
+
+**Drift Monitoring**:
+- **Data Drift**: Monitor input data distributions for shifts over time
+- **Concept Drift**: Monitor relationship between inputs and outputs
+- **Model Performance Drift**: Track accuracy, precision, recall over time
+- **Alert Thresholds**: Trigger retraining if performance drops >5% or drift detected
+
+**Monitoring Frequency**:
+- **Real-Time**: Continuous monitoring of decision distributions
+- **Daily**: Aggregate metrics and drift checks
+- **Weekly**: Bias analysis across demographic groups
+- **Monthly**: Comprehensive model performance review
+
+---
+
+### E. PII Handling and Data Privacy
+
+**Requirement**: All personally identifiable information (PII) must be protected.
+
+**PII Categories**:
+- **Direct Identifiers**: Name, ID number, phone, email, address
+- **Quasi-Identifiers**: Age, gender, nationality, occupation
+- **Sensitive Data**: Medical records, financial details, biometric data
+
+**Protection Measures**:
+1. **Encryption**: PII encrypted at rest and in transit
+2. **Access Control**: Role-based access to PII (need-to-know basis)
+3. **Anonymization**: PII anonymized in training datasets
+4. **Redaction**: PII redacted in logs and reports (except audit logs)
+5. **Retention Limits**: PII deleted after regulatory retention period
+
+**Data Minimization**:
+- Collect only PII necessary for decision-making
+- Avoid storing PII in model artifacts
+- Use pseudonymization where possible
+
+---
+
+### F. Regulatory Alignment
+
+**IFRS Compliance**:
+- **IFRS 17**: Insurance contract accounting standards
+- **Reserve Adequacy**: Models must support reserve calculations
+- **Disclosure Requirements**: Decisions must support financial reporting
+
+**AML Compliance**:
+- **Transaction Monitoring**: Flag suspicious claim patterns
+- **Customer Due Diligence**: Verify claimant identities
+- **Reporting**: Support Suspicious Activity Reports (SARs)
+
+**GCC Regulations**:
+- **Local Insurance Laws**: Comply with country-specific regulations (UAE, KSA, etc.)
+- **Data Residency**: Store data in GCC regions where required
+- **Language Support**: Provide Arabic language support for regulatory filings
+
+**Takaful Principles**:
+- **Sharia Compliance**: Ensure models align with Islamic finance principles
+- **Transparency**: Provide clear explanations compatible with Takaful governance
+- **Fairness**: Avoid interest-based calculations or prohibited practices
+
+---
+
+## Governance Enforcement
+
+### Pre-Deployment Checklist
+Before any model or system goes to production:
+- [ ] Audit logging implemented
+- [ ] Explainability outputs validated
+- [ ] Human override mechanism tested
+- [ ] Bias testing completed and passed
+- [ ] Drift monitoring configured
+- [ ] PII protection measures verified
+- [ ] Regulatory compliance reviewed
+- [ ] Documentation complete
+
+### Production Monitoring
+Once in production:
+- [ ] Daily audit log review
+- [ ] Weekly bias monitoring reports
+- [ ] Monthly drift analysis
+- [ ] Quarterly compliance audits
+- [ ] Annual model revalidation
+
+### Incident Response
+If governance violations detected:
+1. **Immediate**: Pause affected system if critical
+2. **Investigation**: Root cause analysis within 24 hours
+3. **Remediation**: Fix and redeploy within 72 hours
+4. **Reporting**: Notify stakeholders and regulators as required
+5. **Prevention**: Update governance controls to prevent recurrence
+
+---
+
+## Governance Roles and Responsibilities
+
+### Model Owner
+- Ensure model complies with governance standards
+- Maintain model documentation
+- Respond to governance incidents
+
+### Compliance Officer
+- Review audit logs and bias reports
+- Approve model deployments
+- Interface with regulators
+
+### Data Protection Officer
+- Oversee PII handling
+- Ensure data privacy compliance
+- Manage data retention and deletion
+
+### System Administrator
+- Configure audit logging and monitoring
+- Manage access controls
+- Maintain governance infrastructure
+
+---
+
+## Governance Metrics and KPIs
+
+### Decision Quality
+- **Accuracy**: % of decisions that align with ground truth
+- **Precision**: % of positive decisions that are correct
+- **Recall**: % of true positives identified
+- **F1 Score**: Harmonic mean of precision and recall
+
+### Explainability Quality
+- **Completeness**: % of decisions with full explainability outputs
+- **Accuracy**: % of explanations that match actual model behavior
+- **Understandability**: User satisfaction with explanations (survey)
+
+### Fairness
+- **Demographic Parity**: Equal approval rates across groups
+- **Equalized Odds**: Equal true positive and false positive rates across groups
+- **Disparate Impact**: Ratio of approval rates (must be >0.8)
+
+### Compliance
+- **Audit Log Coverage**: % of decisions logged (target: 100%)
+- **PII Protection**: % of PII properly encrypted/anonymized (target: 100%)
+- **Regulatory Violations**: Count of violations (target: 0)
+
+---
+
+## Continuous Improvement
+
+Governance standards are **living documents** that evolve based on:
+- Regulatory changes
+- Lessons learned from incidents
+- Advances in AI governance best practices
+- Stakeholder feedback
+
+**Review Cycle**: Governance standards reviewed and updated quarterly.
+
+---
+
+**BDR Agent Factory** — Unified governance for transparent, accountable, and compliant decision intelligence.

docs/04_EXTENSION_GUIDE.md

@@ -0,0 +1,325 @@
+# Extension Guide — Adding New Capabilities
+
+This document explains **how to add new AI capabilities** to the Bader AI platform while maintaining governance, avoiding duplication, and ensuring consistency.
+
+---
+
+## Core Rules
+
+### Rule 1: Check Before Creating
+**Before adding any new capability**, you must:
+1. Search the **[Capability Dictionary](01_CAPABILITY_DICTIONARY.md)** to verify it doesn't already exist
+2. Check the **[Capability Map](02_CAPABILITY_MAP.md)** to see if an existing capability can be reused
+3. Consult with the platform team to confirm the need for a new capability
+
+**If a capability already exists, you MUST reuse it. No duplicates allowed.**
+
+### Rule 2: Decision-First, Not Capability-First
+New capabilities are only added if they serve a **specific insurance decision**.
+
+**Invalid Reason**: "This is a cool AI technique we should try."  
+**Valid Reason**: "We need this capability to support the 'Should we approve this claim?' decision in ClaimsGPT."
+
+### Rule 3: No Standalone Capability Projects
+Capabilities are **building blocks**, not products.
+
+**Invalid**: Creating a standalone Hugging Face Space for "Anomaly Detection"  
+**Valid**: Adding Anomaly Detection capability to FraudSimulator-AI to support fraud investigation decisions
+
+### Rule 4: Governance from Day One
+Every new capability must comply with **[Governance Standards](03_GOVERNANCE.md)** from the start:
+- Audit logging
+- Explainability
+- Bias testing
+- PII handling
+- Regulatory compliance
+
+---
+
+## Extension Process
+
+### Step 1: Proposal
+Submit a capability proposal with the following information:
+
+```markdown
+## Capability Proposal
+
+**Capability Name**: [e.g., "Geospatial Analysis"]
+
+**Category**: [A-N, e.g., "G - Data, Analytics & Visualization"]
+
+**Definition**: [One-sentence definition of what this capability does]
+
+**Insurance Relevance**: [Why this matters for insurance decisions]
+
+**Decision Use Case**: [Which decision(s) will this capability support?]
+
+**System Integration**: [Which Space(s) will use this capability?]
+
+**Model**: [Which model will implement this capability?]
+
+**Dataset**: [Which dataset will support this capability?]
+
+**Governance Considerations**: [Any special governance requirements?]
+
+**Alternatives Considered**: [Why can't existing capabilities be used?]
+```
+
+**Example**:
+```markdown
+## Capability Proposal
+
+**Capability Name**: Geospatial Analysis
+
+**Category**: G - Data, Analytics & Visualization
+
+**Definition**: Analyzing geographic data to identify spatial patterns and relationships.
+
+**Insurance Relevance**: Detect fraud rings operating in specific geographic areas, assess catastrophe risk by location, optimize claims routing by proximity.
+
+**Decision Use Case**: 
+- FraudSimulator-AI: "Should this claim be investigated?" (detect geographic fraud clusters)
+- AutoRiskScoreEngine: "What risk band does this policy belong to?" (assess location-based risk)
+
+**System Integration**: 
+- FraudSimulator-AI
+- AutoRiskScoreEngine
+
+**Model**: 
+- fraud-risk-agent (add geospatial features)
+- underwriting-risk-agent (add location risk scoring)
+
+**Dataset**: 
+- fraud-simulator-dataset (add geographic coordinates)
+- underwriting-risk-dataset (add location data)
+
+**Governance Considerations**: 
+- Location data may be PII (requires anonymization)
+- Must avoid geographic bias (e.g., unfairly flagging certain neighborhoods)
+
+**Alternatives Considered**: 
+- Existing "Anomaly Detection" capability does not account for spatial relationships
+- Existing "Risk Scoring" does not incorporate geographic clustering
+```
+
+### Step 2: Review and Approval
+The proposal is reviewed by:
+1. **Platform Team**: Technical feasibility and architecture fit
+2. **Compliance Officer**: Governance and regulatory alignment
+3. **Business Owner**: Decision relevance and business value
+
+**Approval Criteria**:
+- [ ] Capability does not duplicate existing capabilities
+- [ ] Capability serves a specific insurance decision
+- [ ] Capability can be integrated into existing systems
+- [ ] Governance requirements are understood and feasible
+- [ ] Business value is clear
+
+### Step 3: Implementation
+Once approved, implement the capability following these steps:
+
+#### 3.1 Update Capability Dictionary
+Add the new capability to **[01_CAPABILITY_DICTIONARY.md](01_CAPABILITY_DICTIONARY.md)** under the appropriate category:
+
+```markdown
+### Geospatial Analysis
+**Definition**: Analyzing geographic data to identify spatial patterns and relationships.  
+**Insurance Relevance**: Detect fraud rings in specific areas, assess catastrophe risk by location, optimize claims routing.
+```
+
+#### 3.2 Update Capability Map
+Add mappings to **[02_CAPABILITY_MAP.md](02_CAPABILITY_MAP.md)**:
+
+```markdown
+| Geospatial Analysis | G - Analytics | FraudSimulator-AI, AutoRiskScoreEngine | fraud-risk-agent, underwriting-risk-agent | fraud-simulator-dataset, underwriting-risk-dataset | Detect geographic fraud clusters, assess location risk |
+```
+
+#### 3.3 Implement in Model
+Add the capability to the relevant model(s):
+- Update model code to include the new capability
+- Update model contract to include new outputs (if applicable)
+- Version the model (semantic versioning: major.minor.patch)
+
+#### 3.4 Update Dataset
+Ensure the dataset supports the new capability:
+- Add required features (e.g., geographic coordinates)
+- Validate data quality
+- Update dataset documentation
+
+#### 3.5 Implement Governance
+Ensure the capability complies with governance standards:
+- [ ] Audit logging for decisions using this capability
+- [ ] Explainability outputs (how does this capability influence decisions?)
+- [ ] Bias testing (does this capability introduce geographic bias?)
+- [ ] PII handling (is location data properly anonymized?)
+- [ ] Drift monitoring (track performance over time)
+
+#### 3.6 Update System Documentation
+Update the relevant Space's README and documentation:
+- List the new capability
+- Explain how it's used in decision-making
+- Provide examples
+
+### Step 4: Testing and Validation
+Before deploying to production:
+
+#### 4.1 Unit Tests
+- Test the capability in isolation
+- Verify outputs match expected behavior
+
+#### 4.2 Integration Tests
+- Test the capability within the full system
+- Verify it integrates correctly with other capabilities
+
+#### 4.3 Governance Tests
+- Verify audit logging works
+- Verify explainability outputs are generated
+- Run bias tests
+- Validate PII handling
+
+#### 4.4 Performance Tests
+- Measure latency impact
+- Ensure scalability
+
+#### 4.5 User Acceptance Testing
+- Test with real users (claims handlers, underwriters, etc.)
+- Gather feedback on explainability and usability
+
+### Step 5: Deployment
+Deploy the capability following the standard deployment process:
+1. Deploy to staging environment
+2. Run smoke tests
+3. Deploy to production with monitoring
+4. Monitor for issues in first 48 hours
+
+### Step 6: Documentation and Communication
+After deployment:
+1. Update all documentation (Capability Dictionary, Capability Map, System READMEs)
+2. Communicate the new capability to stakeholders
+3. Provide training if needed
+4. Update governance dashboards
+
+---
+
+## Versioning Rules
+
+### Capability Versioning
+Capabilities themselves are not versioned. However, their **implementations** in models are versioned.
+
+### Model Versioning
+Use semantic versioning: `MAJOR.MINOR.PATCH`
+
+- **MAJOR**: Breaking changes (e.g., model contract changes)
+- **MINOR**: New capabilities added (backward compatible)
+- **PATCH**: Bug fixes, performance improvements
+
+**Example**:
+- `fraud-risk-agent v1.0.0`: Initial release
+- `fraud-risk-agent v1.1.0`: Added Geospatial Analysis capability
+- `fraud-risk-agent v1.1.1`: Fixed bug in Geospatial Analysis
+- `fraud-risk-agent v2.0.0`: Changed model contract (breaking change)
+
+### Dataset Versioning
+Datasets are versioned when:
+- New features are added
+- Data quality issues are fixed
+- New data is added
+
+Use date-based versioning: `YYYY-MM-DD` or semantic versioning.
+
+---
+
+## Deprecation Process
+
+If a capability is no longer needed:
+
+### Step 1: Deprecation Notice
+- Mark the capability as deprecated in the Capability Dictionary
+- Provide a deprecation timeline (e.g., 90 days)
+- Suggest alternatives
+
+### Step 2: Migration
+- Update systems to use alternative capabilities
+- Test thoroughly
+
+### Step 3: Removal
+- Remove the capability from the Capability Dictionary and Capability Map
+- Archive related code and documentation
+- Update all affected systems
+
+---
+
+## Common Pitfalls to Avoid
+
+### ❌ Pitfall 1: Creating Duplicate Capabilities
+**Problem**: Adding "Fraud Pattern Detection" when "Anomaly Detection" already exists.  
+**Solution**: Reuse "Anomaly Detection" and configure it for fraud patterns.
+
+### ❌ Pitfall 2: Adding Capabilities Without Decision Context
+**Problem**: Adding "Sentiment Analysis" because it's trendy, without a clear decision use case.  
+**Solution**: Only add capabilities that serve specific insurance decisions.
+
+### ❌ Pitfall 3: Ignoring Governance
+**Problem**: Deploying a new capability without bias testing or explainability.  
+**Solution**: Follow the governance checklist before deployment.
+
+### ❌ Pitfall 4: Creating Standalone Capability Spaces
+**Problem**: Creating a separate Hugging Face Space for "Risk Scoring."  
+**Solution**: Integrate "Risk Scoring" into FraudSimulator-AI and AutoRiskScoreEngine.
+
+### ❌ Pitfall 5: Poor Documentation
+**Problem**: Adding a capability but not updating the Capability Dictionary or Map.  
+**Solution**: Documentation is mandatory, not optional.
+
+---
+
+## Extension Checklist
+
+Use this checklist when adding a new capability:
+
+### Pre-Implementation
+- [ ] Searched Capability Dictionary for duplicates
+- [ ] Verified capability serves a specific decision
+- [ ] Submitted capability proposal
+- [ ] Received approval from Platform Team, Compliance, and Business Owner
+
+### Implementation
+- [ ] Updated Capability Dictionary
+- [ ] Updated Capability Map
+- [ ] Implemented in model(s)
+- [ ] Updated dataset(s)
+- [ ] Implemented governance (audit logging, explainability, bias testing, PII handling)
+- [ ] Updated system documentation
+
+### Testing
+- [ ] Unit tests passed
+- [ ] Integration tests passed
+- [ ] Governance tests passed
+- [ ] Performance tests passed
+- [ ] User acceptance testing completed
+
+### Deployment
+- [ ] Deployed to staging
+- [ ] Smoke tests passed
+- [ ] Deployed to production
+- [ ] Monitoring configured
+- [ ] No critical issues in first 48 hours
+
+### Post-Deployment
+- [ ] Documentation finalized
+- [ ] Stakeholders notified
+- [ ] Training provided (if needed)
+- [ ] Governance dashboards updated
+
+---
+
+## Contact and Support
+
+For questions about adding new capabilities:
+- **Platform Team**: platform@bdr-ai.org
+- **Compliance Officer**: compliance@bdr-ai.org
+- **Documentation**: [BDR Agent Factory](https://huggingface.co/spaces/bdr-ai-org/BDR-Agent-Factory)
+
+---
+
+**BDR Agent Factory** — Structured, governed, and decision-driven capability extension.

requirements.txt

@@ -0,0 +1 @@
+gradio==4.44.0

ui/capability_browser.py

@@ -0,0 +1,166 @@
+import gradio as gr
+import json
+import os
+
+# Load capability registry
+def load_registry():
+    registry_path = os.path.join(os.path.dirname(__file__), '..', 'data', 'capability_registry.json')
+    with open(registry_path, 'r') as f:
+        return json.load(f)
+
+registry = load_registry()
+capabilities = registry['capabilities']
+
+# Extract unique values for filters
+all_categories = sorted(list(set([cap['category'] for cap in capabilities])))
+all_spaces = sorted(list(set([space for cap in capabilities for space in cap['used_in_spaces']])))
+all_models = sorted(list(set([model for cap in capabilities for model in cap['models']])))
+all_datasets = sorted(list(set([dataset for cap in capabilities for dataset in cap['datasets']])))
+
+def search_capabilities(search_text, category_filter, space_filter, model_filter, dataset_filter):
+    """Search and filter capabilities based on user inputs"""
+    filtered_caps = capabilities
+    
+    # Apply search text filter
+    if search_text:
+        search_lower = search_text.lower()
+        filtered_caps = [
+            cap for cap in filtered_caps 
+            if search_lower in cap['capability_name'].lower() 
+            or search_lower in cap['category'].lower()
+            or search_lower in ' '.join(cap['insurance_decisions']).lower()
+        ]
+    
+    # Apply category filter
+    if category_filter and category_filter != "All":
+        filtered_caps = [cap for cap in filtered_caps if cap['category'] == category_filter]
+    
+    # Apply space filter
+    if space_filter and space_filter != "All":
+        filtered_caps = [cap for cap in filtered_caps if space_filter in cap['used_in_spaces']]
+    
+    # Apply model filter
+    if model_filter and model_filter != "All":
+        filtered_caps = [cap for cap in filtered_caps if model_filter in cap['models']]
+    
+    # Apply dataset filter
+    if dataset_filter and dataset_filter != "All":
+        filtered_caps = [cap for cap in filtered_caps if dataset_filter in cap['datasets']]
+    
+    # Generate HTML output
+    if not filtered_caps:
+        return "<div style='padding: 20px; text-align: center;'><h3>No capabilities found matching your criteria.</h3></div>"
+    
+    html_output = f"<div style='padding: 20px;'><h3>Found {len(filtered_caps)} Capability(ies)</h3>"
+    
+    for cap in filtered_caps:
+        html_output += f"""
+        <div style='border: 1px solid #ddd; border-radius: 8px; padding: 20px; margin: 15px 0; background: #f9f9f9;'>
+            <h3 style='margin-top: 0; color: #2c3e50;'>🔹 {cap['capability_name']}</h3>
+            <p><strong>Category:</strong> {cap['category']}</p>
+            <p><strong>Used in Spaces:</strong> {', '.join(cap['used_in_spaces'])}</p>
+            <p><strong>Models:</strong> {', '.join(cap['models'])}</p>
+            <p><strong>Datasets:</strong> {', '.join(cap['datasets'])}</p>
+            <p><strong>Insurance Decisions:</strong> {', '.join(cap['insurance_decisions'])}</p>
+            <p><strong>Governance Required:</strong> {'✅ Yes' if cap['governance_required'] else '❌ No'}</p>
+        </div>
+        """
+    
+    html_output += "</div>"
+    return html_output
+
+def create_capability_browser():
+    """Create the Gradio interface for capability browsing"""
+    
+    with gr.Blocks(title="BDR Agent Factory - Capability Browser", theme=gr.themes.Soft()) as demo:
+        gr.Markdown("""
+        # 🏭 BDR Agent Factory — Capability Browser
+        
+        **Enterprise Decision Intelligence Architecture for Insurance & Regulated Markets**
+        
+        Browse and search AI capabilities used across the Bader AI platform. Filter by category, system, model, or dataset to understand how capabilities map to insurance decisions.
+        """)
+        
+        with gr.Row():
+            with gr.Column(scale=2):
+                search_box = gr.Textbox(
+                    label="🔍 Search Capabilities",
+                    placeholder="Search by capability name, category, or decision...",
+                    lines=1
+                )
+            
+        with gr.Row():
+            category_dropdown = gr.Dropdown(
+                choices=["All"] + all_categories,
+                value="All",
+                label="📂 Filter by Category",
+                interactive=True
+            )
+            space_dropdown = gr.Dropdown(
+                choices=["All"] + all_spaces,
+                value="All",
+                label="🚀 Filter by Space",
+                interactive=True
+            )
+            model_dropdown = gr.Dropdown(
+                choices=["All"] + all_models,
+                value="All",
+                label="🤖 Filter by Model",
+                interactive=True
+            )
+            dataset_dropdown = gr.Dropdown(
+                choices=["All"] + all_datasets,
+                value="All",
+                label="📊 Filter by Dataset",
+                interactive=True
+            )
+        
+        search_button = gr.Button("Search", variant="primary", size="lg")
+        
+        results_html = gr.HTML(
+            value=search_capabilities("", "All", "All", "All", "All"),
+            label="Results"
+        )
+        
+        # Set up event handlers
+        search_button.click(
+            fn=search_capabilities,
+            inputs=[search_box, category_dropdown, space_dropdown, model_dropdown, dataset_dropdown],
+            outputs=results_html
+        )
+        
+        # Also trigger search on Enter key in search box
+        search_box.submit(
+            fn=search_capabilities,
+            inputs=[search_box, category_dropdown, space_dropdown, model_dropdown, dataset_dropdown],
+            outputs=results_html
+        )
+        
+        gr.Markdown("""
+        ---
+        
+        ## 📚 Documentation
+        
+        - **[Overview](https://huggingface.co/spaces/bdr-ai-org/BDR-Agent-Factory/blob/main/docs/00_OVERVIEW.md)**: What is BDR Agent Factory?
+        - **[Capability Dictionary](https://huggingface.co/spaces/bdr-ai-org/BDR-Agent-Factory/blob/main/docs/01_CAPABILITY_DICTIONARY.md)**: Full catalog of AI capabilities (A–N)
+        - **[Capability Map](https://huggingface.co/spaces/bdr-ai-org/BDR-Agent-Factory/blob/main/docs/02_CAPABILITY_MAP.md)**: Mappings to systems, models, datasets, decisions
+        - **[Governance Standards](https://huggingface.co/spaces/bdr-ai-org/BDR-Agent-Factory/blob/main/docs/03_GOVERNANCE.md)**: Unified governance policies
+        - **[Extension Guide](https://huggingface.co/spaces/bdr-ai-org/BDR-Agent-Factory/blob/main/docs/04_EXTENSION_GUIDE.md)**: How to add new capabilities
+        
+        ## 🔗 Linked Systems
+        
+        - **[ClaimsGPT](https://huggingface.co/spaces/bdr-ai-org/ClaimsGPT)**: AI-powered claim decision intelligence
+        - **[FraudSimulator-AI](https://huggingface.co/spaces/bdr-ai-org/FraudSimulator-AI)**: Fraud risk and anomaly detection
+        - **[AutoRiskScoreEngine](https://huggingface.co/spaces/bdr-ai-org/AutoRiskScoreEngine)**: IFRS-ready underwriting risk assessment
+        - **[InsuranceKnowledgeAgent](https://huggingface.co/spaces/bdr-ai-org/InsuranceKnowledgeAgent)**: RAG-powered policy knowledge
+        
+        ---
+        
+        **BDR Agent Factory** — The authoritative capability registry for Bader AI, the GCC Insurance Decision Intelligence Platform.
+        """)
+    
+    return demo
+
+if __name__ == "__main__":
+    demo = create_capability_browser()
+    demo.launch()