Added new features

Author: Vish Devarajan

BENCHMARKS.md

@@ -34,6 +34,7 @@ Operational questions this should answer:
 - What would have been blocked this week?
 - Which routes trigger the most prompt-injection findings?
 - Which features or workflows are producing the most noise?
+- Which users or identity providers are associated with the most risky traffic?
 - Which tenants or models show the highest concentration of severe findings?
 - Which rules are most common before enforcement is turned on?
 - Did the latest release increase alerts on the same route mix?

CHANGELOG.md

@@ -1,5 +1,19 @@
 # Changelog
 
+## 0.2.0
+
+- Added `ValueAtRiskCircuitBreaker` for high-value tool/action thresholds with session revocation and MFA-style escalation flags
+- Added `ShadowConsensusAuditor` integration in tool gating for logic-conflict review on high-impact actions
+- Added `DigitalTwinOrchestrator` for mock tool sandboxes and pre-production twin testing
+- Added `suggest_policy_override()` for self-healing policy tuning suggestions after approved false positives
+- Added signed agent passports in `AgentIdentityRegistry`
+
+## 0.1.9
+
+- Added enterprise telemetry enrichment with SSO/user attribution on emitted events and audit records
+- Added Power BI-friendly record builders and exporter hooks for telemetry pipelines
+- Expanded operational summaries to break down findings by user and identity provider
+
 ## 0.1.8
 
 - Expanded enterprise rollout guidance for controlled pilots, internal shield wrappers, and false-positive tuning

MIGRATING.md

@@ -1,4 +1,4 @@
-# Migrating to 0.1.8
+# Migrating to 0.1.9
 
 ## Stable Contracts
 
@@ -21,6 +21,12 @@ These contracts are also exposed in `CORE_INTERFACES` so applications can log or
 - Added custom prompt detector hooks for domain tuning
 - Expanded rollout guidance, benchmarks, and regression notes
 
+## What Changed in 0.1.9
+
+- Added identity-aware telemetry enrichment for SSO-backed applications
+- Added Power BI-friendly export helpers and telemetry exporter hooks
+- Expanded summaries to support user-level and identity-provider reporting
+
 ## What Changed in 0.1.8
 
 - Added explicit guidance for controlled-pilot rollout and internal wrapper adoption

README.md

@@ -77,6 +77,8 @@ The current recommendation for enterprise teams is a controlled pilot first: sta
 
 Use `summarize_operational_telemetry()` with emitted telemetry events when you want route-level, tenant-level, and model-level summaries, blocked-event counts, and rollout visibility for operators.
 
+Enterprise deployments can also enrich emitted events with SSO/user context and forward flattened records to Power BI or other downstream reporting systems.
+
 ### Output grounding and tone review
 
 `OutputFirewall` can compare a response to retrieval documents and flag unsupported claims or unprofessional tone before the answer leaves your service.
@@ -105,6 +107,8 @@ Protects the response path by checking outputs for secret leaks, unsafe code pat
 
 Protects tool execution with allowlists, blocklists, validators, and approval-required workflows.
 
+It can also integrate with `ValueAtRiskCircuitBreaker` for high-value actions and `ShadowConsensusAuditor` for secondary logic review before sensitive tools execute.
+
 ### `RetrievalSanitizer`
 
 Helps keep hostile or manipulative text in retrieved documents from becoming model instructions.
@@ -170,6 +174,48 @@ def create_model_shield(shield):
     return run
 ```
 
+## Add SSO-aware Telemetry and Power BI Export
+
+```python
+from blackwall_llm_shield import BlackwallShield, PowerBIExporter
+
+shield = BlackwallShield(
+    identity_resolver=lambda metadata: {
+        "user_id": ((metadata.get("sso") or {}).get("subject")),
+        "user_email": ((metadata.get("sso") or {}).get("email")),
+        "user_name": ((metadata.get("sso") or {}).get("displayName")),
+        "identity_provider": ((metadata.get("sso") or {}).get("provider")),
+        "groups": ((metadata.get("sso") or {}).get("groups") or []),
+    },
+    telemetry_exporters=[
+        PowerBIExporter(endpoint_url="https://example.powerbi.local/push"),
+    ],
+)
+```
+
+## Protect High-value Actions with a VaR Breaker and Consensus Auditor
+
+```python
+firewall = ToolPermissionFirewall(
+    allowed_tools=["issue_refund"],
+    value_at_risk_circuit_breaker=ValueAtRiskCircuitBreaker(max_value_per_window=5000),
+    consensus_auditor=ShadowConsensusAuditor(),
+    consensus_required_for=["issue_refund"],
+)
+```
+
+## Generate a Digital Twin for Sandbox Testing
+
+```python
+twin = DigitalTwinOrchestrator(
+    tool_schemas=[
+        {"name": "lookup_order", "mock_response": {"order_id": "ord_1", "status": "mocked"}},
+    ]
+).generate()
+
+twin["simulate_call"]("lookup_order", {"order_id": "ord_1"})
+```
+
 ## Strict JSON Workflow Pattern
 
 ```python
@@ -288,6 +334,8 @@ shield = BlackwallShield(
 summary = summarize_operational_telemetry(events)
 print(summary["by_route"])
 print(summary["by_feature"])
+print(summary["by_user"])
+print(summary["by_identity_provider"])
 print(summary["noisiest_routes"])
 print(summary["weekly_block_estimate"])
 print(summary["by_tenant"])
@@ -299,6 +347,14 @@ print(summary["highest_severity"])
 
 Produces signed events you can summarize into operations dashboards or audit pipelines.
 
+## Advanced Agent Controls
+
+- `ValueAtRiskCircuitBreaker` for financial or high-value operational actions
+- `ShadowConsensusAuditor` for second-model or secondary-review logic conflict checks
+- `DigitalTwinOrchestrator` for mock tool environments and sandbox simulations
+- `suggest_policy_override()` for narrow false-positive tuning suggestions after HITL approvals
+- `AgentIdentityRegistry.issue_signed_passport()` for signed agent identity exchange
+
 ## Included Examples
 
 - [`examples/python-fastapi/main.py`](/Users/vishnu/Documents/blackwall-llm-shield/blackwall-llm-shield-python/examples/python-fastapi/main.py)
@@ -339,6 +395,8 @@ For Gemini-heavy apps, the bundled adapter now preserves system instructions plu
 - A controlled pilot is a good fit today when you want shadow-mode prompt and output protection without forcing hard blocking on every route immediately.
 - If you prefer not to depend on Blackwall directly everywhere, wrap it behind your own internal model-security abstraction and expose only the contract your app teams need.
 - For broader approval, focus rollout reviews on false-positive rates, noisiest routes, and latency budgets alongside jailbreak coverage.
+- For executive or staff-facing workflows, always attach authenticated identity metadata so telemetry can answer which user triggered which risky request or output event.
+- For high-impact agentic workflows, combine tool approval, VaR limits, digital-twin tests, and signed agent passports instead of relying on a single detector.
 
 ## Rollout Notes
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "vpdeva-blackwall-llm-shield-python"
-version = "0.1.8"
+version = "0.2.0"
 description = "Open-source Python enterprise LLM protection toolkit for Python services"
 readme = "README.md"
 requires-python = ">=3.9"

src/blackwall_llm_shield/__init__.py

@@ -5,22 +5,27 @@
     BlackwallFastAPIMiddleware,
     BlackwallShield,
     CoTScanner,
+    DigitalTwinOrchestrator,
     ImageMetadataScanner,
     LightweightIntentScorer,
     MCPSecurityProxy,
     OutputFirewall,
     RetrievalSanitizer,
     ShadowAIDiscovery,
     SessionBuffer,
+    ShadowConsensusAuditor,
     TokenBudgetFirewall,
     ToolPermissionFirewall,
+    ValueAtRiskCircuitBreaker,
     VisualInstructionDetector,
     SHIELD_PRESETS,
     CORE_INTERFACES,
     POLICY_PACKS,
     build_shield_options,
     summarize_operational_telemetry,
     build_admin_dashboard_model,
+    build_enterprise_telemetry_event,
+    build_powerbi_record,
     create_fastapi_guard,
     create_langchain_callbacks,
     create_flask_middleware,
@@ -39,7 +44,10 @@
     mask_text,
     mask_value,
     normalize_messages,
+    normalize_identity_metadata,
     parse_json_output,
+    PowerBIExporter,
+    suggest_policy_override,
     run_red_team_suite,
     rehydrate_from_bundle,
     rehydrate_response,
@@ -72,6 +80,7 @@
     "BlackwallShield",
     "CoTScanner",
     "CORE_INTERFACES",
+    "DigitalTwinOrchestrator",
     "FastTextIntentScorer",
     "ImageMetadataScanner",
     "LightweightIntentScorer",
@@ -80,15 +89,19 @@
     "RetrievalSanitizer",
     "ShadowAIDiscovery",
     "SessionBuffer",
+    "ShadowConsensusAuditor",
     "TokenBudgetFirewall",
     "ToolPermissionFirewall",
+    "ValueAtRiskCircuitBreaker",
     "VisualInstructionDetector",
     "POLICY_PACKS",
     "ProviderAdapter",
     "SHIELD_PRESETS",
     "build_shield_options",
     "summarize_operational_telemetry",
     "build_admin_dashboard_model",
+    "build_enterprise_telemetry_event",
+    "build_powerbi_record",
     "create_fastapi_guard",
     "create_langchain_callbacks",
     "create_flask_middleware",
@@ -108,7 +121,10 @@
     "mask_text",
     "mask_value",
     "normalize_messages",
+    "normalize_identity_metadata",
     "parse_json_output",
+    "PowerBIExporter",
+    "suggest_policy_override",
     "run_red_team_suite",
     "rehydrate_from_bundle",
     "rehydrate_response",