Feature/sentinel kql #743

SinsBre · 2025-09-23T08:59:01Z

🎯 Overview

This PR introduces a new plugin for Microsoft Sentinel (Azure Log
Analytics) that enables users to query security data using KQL (Kusto
Query Language) and create graph visualizations directly from their
Sentinel workspaces.

✨ What's New

Core Features

New SentinelMixin plugin - Query Microsoft Sentinel Log Analytics
workspaces using KQL
Multiple authentication methods:
- Device code authentication (interactive with code/URL)
- Service Principal (client_id, client_secret, tenant_id)
- DefaultAzureCredential (Azure CLI, Managed Identity, Environment)
KQL query execution with automatic DataFrame conversion
Nested data handling - Automatic unwrapping of complex JSON columns
common in Sentinel
Helper methods for workspace exploration:
- sentinel_tables() - List available tables
- sentinel_schema() - Get table schemas
- kql_last() - Convenience method for recent data queries

Integration

Seamlessly integrated into the Graphistry ecosystem alongside existing
Kusto and Spanner plugins
Full module-level access via graphistry.configure_sentinel()
Follows established plugin patterns for consistency

📝 Changes

New Files

graphistry/plugins/sentinel.py - Main plugin implementation (~650 lines)
graphistry/plugins_types/sentinel_types.py - Type definitions and data
structures
graphistry/tests/test_sentinel.py - Comprehensive unit tests (379 lines)
demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis
.ipynb - Example notebook
demos/demos_databases_apis/microsoft/sentinel/example.env - Environment
template

Modified Files

graphistry/plotter.py - Added SentinelMixin to Plotter class
graphistry/pygraphistry.py - Added module-level exports
graphistry/init.py - Added sentinel methods to public API
setup.py - Added dependencies: azure-monitor-query>=1.2.0,
azure-identity>=1.12.0

🔒 Security Analysis Use Cases

The notebook demonstrates real-world security analysis scenarios:

Failed login detection and analysis
Security alert correlation
Network traffic analysis
User-IP relationship mapping
Alert-entity correlation graphs

🧪 Testing

Comprehensive unit tests with mocked Azure responses
Test coverage for all authentication methods
Nested data unwrapping tests
Error handling validation

📚 Documentation

Full docstrings with examples for all public methods
Security-focused example notebook with common KQL patterns
Environment setup guide with .env template
Troubleshooting tips and best practices

🔑 Authentication

Supports multiple authentication patterns for different use cases:

Device code (interactive)

g = graphistry.configure_sentinel(workspace_id="...",
use_device_auth=True)

Service Principal (production)

g = graphistry.configure_sentinel(
workspace_id="...",
tenant_id="...",
client_id="...",
client_secret="..."
)

DefaultAzureCredential (flexible)

g = graphistry.configure_sentinel(workspace_id="...")

📊 Example Usage

import graphistry
from datetime import timedelta

Configure connection

g = graphistry.configure_sentinel(
workspace_id="your-workspace-id",
use_device_auth=True
)

Query failed logins

failed_logins = g.kql("""
SigninLogs
| where ResultType != "0"
| summarize Count=count() by UserPrincipalName, IPAddress
| top 20 by Count
""", timespan=timedelta(days=7))

Create graph visualization

graph = g.nodes(users).edges(connections)
.encode_point_color('node_type')
.plot()

🚀 Benefits

Unified security analytics - Combine Sentinel's powerful KQL with
Graphistry's graph visualization
No data movement - Query data directly from Sentinel workspaces
Enterprise ready - Multiple auth methods including managed identity
support
Developer friendly - Follows existing Kusto plugin patterns for
familiarity

🔄 Compatibility

Python 3.8+
Compatible with latest Azure Monitor Query SDK
Works with all Log Analytics workspace versions
Handles both legacy and modern column formats

📈 Impact

This connector enables security teams to:

Visualize complex attack patterns as graphs
Identify lateral movement and privilege escalation
Correlate alerts across multiple data sources
Build visual threat hunting workflows

✅ Checklist

Code follows project style guidelines
Unit tests pass
Documentation updated
Example notebook provided
Dependencies added to setup.py
Type hints included
Error handling implemented
Module exports configured

Breaking Changes: None

Migration Guide: Not applicable (new feature)

Related Issues: Addresses need for Microsoft Sentinel integration

Testing Instructions:

Install with pip install -e .[sentinel]
Set up Azure authentication (device code or service principal)
Run the example notebook with your workspace ID
Verify KQL queries return DataFrames
Test graph visualizations work correctly

- Add SentinelConfig dataclass for connection configuration - Add SentinelQueryResult class for query results - Add custom exceptions for connection and query errors - Support for Azure authentication credentials

- Add SentinelMixin class extending Plottable - Implement configure_sentinel() with multiple auth methods - Support Azure CLI, Service Principal, and custom credentials - Add sentinel_from_client() for existing client reuse - Implement health check and basic query infrastructure - Add client initialization with DefaultAzureCredential support

- Implement kql() method with timespan and nested data support - Add kql_last() convenience method for recent data queries - Add sentinel_tables() and sentinel_schema() helper methods - Port nested data unwrapping from Kusto plugin - Support for multiple table responses - Handle JSON strings and dynamic columns in Sentinel results

- Add SentinelMixin import to plotter.py - Include SentinelMixin in Plotter class hierarchy - Update documentation to list Sentinel integration

- Add azure-monitor-query>=1.2.0 and azure-identity>=1.12.0 to sentinel extras - Add 'Sentinel' to package keywords for discoverability

- Test configuration methods (basic, service principal, custom credential) - Test KQL query execution (single/multiple tables, timespan) - Test helper methods (kql_last, sentinel_tables, sentinel_schema) - Test health check functionality - Test nested data unwrapping and JSON parsing - Test authentication initialization flows - Add mock-based testing to avoid API dependencies

- Demonstrate Azure CLI and Service Principal authentication - Show security use cases: failed logins, alerts, network analysis - Include graph visualizations for user-IP and alert correlations - Provide examples of multi-table KQL queries - Cover workspace exploration and schema inspection - Add troubleshooting guidance and next steps

- Replace hardcoded credentials with environment variables - Add example.env template file - Support custom .env file locations - Include python-dotenv dependency instructions - Improve security by avoiding credential commits

- Replace username/password with personal_key_id/personal_key_secret - Update example.env with new credential format - Use modern Graphistry authentication method - Maintain security with environment variables

- Add configure_sentinel and sentinel_from_client to __init__.py - Add corresponding methods in pygraphistry.py - Fix module-level access to Sentinel functionality

- Add configure_sentinel and sentinel_from_client assignments - Follow same pattern as other plugins (Kusto, Spanner) - Enable direct import from graphistry module

- Add use_device_auth parameter to configure_sentinel() - Support DeviceCodeCredential for interactive authentication - Show code and URL for authentication like Kusto plugin - Update type definitions and method signatures - Provide authentication precedence documentation

- Support both object columns (with .name/.type) and string columns - Default to 'string' type when type info not available - Handle missing table.name attribute gracefully - Fix AttributeError in query response processing

- Replace union withsource=TableName query that caused conflicts - Use Usage table to get DataType (table names) instead - Extend timespan to 30 days for better table coverage - Avoid SEM0001 semantic error from existing TableName columns

- Fix references from TableName to DataType column - Clean up notebook code for consistent formatting - Add device authentication example - Ensure table listing and schema queries work correctly

- Replace bind() with encode_point_color() and encode_edge_color() - Use encode_edge_size() for edge weight visualization - Fix method signatures to match Graphistry API

- Add better error handling and user guidance - Include troubleshooting tips and alternative auth methods - Improve error messages and empty result handling - Add comprehensive summary and next steps - Make notebook more robust for different workspace configurations

- Import SentinelMixin to fix F821 undefined name errors - Resolves flake8 lint issues in test_sentinel.py

- Fix W504: Move binary operator before line break - Fix W292: Add newlines at end of all files - Fix F841: Remove unused local variables in tests - All flake8 lint issues now resolved

…KustoMixin

lmeyerov · 2025-09-24T16:46:34Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+    "\n",
+    "## Getting Started\n",
+    "\n",
+    "### Option 1: Azure CLI Authentication (Recommended for Development)\n",


add an azure cli auth link to here + option 2?

lmeyerov · 2025-09-24T16:47:39Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": "# Alternative: Service Principal authentication from .env file\n# Uncomment the lines below if you prefer Service Principal over device authentication\n# g = graphistry.configure_sentinel(\n#     workspace_id=os.getenv('SENTINEL_WORKSPACE_ID'),\n#     tenant_id=os.getenv('AZURE_TENANT_ID'),\n#     client_id=os.getenv('AZURE_CLIENT_ID'),\n#     client_secret=os.getenv('AZURE_CLIENT_SECRET')\n# )\n\n# Alternative: Use DefaultAzureCredential (tries Azure CLI, Managed Identity, etc.)\n# g = graphistry.configure_sentinel(\n#     workspace_id=os.getenv('SENTINEL_WORKSPACE_ID')\n# )"


something i like to do is

if False: .... else: print('Skipped, switch if statement to enable')

That way they don't have to edit all the #'s

lmeyerov · 2025-09-24T16:50:45Z

graphistry/plugins/sentinel.py

+                g.sentinel_health_check()  # Verify connection works
+        """
+        try:
+            self._sentinel_query("Heartbeat | take 1", timespan=timedelta(hours=1))


chatgpt thinks this is unsafe as it assumes azure log analytics, while union withsource=TableName * | take 1 can work on sentinel without

lmeyerov · 2025-09-24T16:52:53Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+    "\n",
+    "1. **Azure Access**: You need access to a Microsoft Sentinel workspace\n",
+    "2. **Authentication**: Either Azure CLI (`az login`) or service principal credentials\n",
+    "3. **Dependencies**: Install required packages\n",


Add 4.a..4.z with assumptions about Sentinel setup, and if they don't satisfy any individual, rec workarounds?

lmeyerov · 2025-09-24T16:53:08Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+    "## Prerequisites\n",
+    "\n",
+    "1. **Azure Access**: You need access to a Microsoft Sentinel workspace\n",
+    "2. **Authentication**: Either Azure CLI (`az login`) or service principal credentials\n",


add docs link?

and if az login mode, do we need to qualify az login running in same server/container as the notebook, or ?

lmeyerov · 2025-09-24T16:54:43Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+   "source": [
+    "# Microsoft Sentinel Security Analysis with Graphistry\n",
+    "\n",
+    "This notebook demonstrates how to use Graphistry with Microsoft Sentinel (Log Analytics) to perform security analysis and visualization using KQL queries.\n",


qualify which visualizations we'll do, like ... graph visualization using KQL queries for logins, alerts, traffic, and other built-in sentinel tables

lmeyerov · 2025-09-24T16:59:54Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+   "source": [
+    "## Security Analysis Examples\n",
+    "\n",
+    "### 1. Failed Login Analysis"


Instead of a separate analysis section without graphistry viz, I would instead fold Graph Visualization into Security Analysis Examples so just 1 section of use cases, where each is like

## Security Analysis Examples ### Failed logins: IP <> Principal <> Device graph #### 1. Analysis ... #### 2. Visualization ### Next use case ...

All of the use cases sound relevant

lmeyerov · 2025-09-24T17:00:10Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Security Analysis Examples\n",


As there are a lot of use cases, add a bulleted list enumerating which will be here?

lmeyerov · 2025-09-24T17:02:55Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": "## Summary\n\nThis notebook demonstrated:\n\n1. **Connecting to Microsoft Sentinel** using Azure authentication (device code, service principal, or DefaultAzureCredential)\n2. **Exploring available data** with `sentinel_tables()` and `sentinel_schema()`\n3. **Security analysis** using KQL queries for:\n   - Failed login analysis\n   - Security alerts monitoring\n   - Network traffic analysis\n4. **Graph visualization** of:\n   - User-IP relationships\n   - Alert correlations\n5. **Advanced correlation** across multiple data sources\n\n## Next Steps\n\n- **Customize queries** for your specific security use cases and available data tables\n- **Create automated dashboards** by scheduling notebook execution\n- **Integrate with threat intelligence** feeds using additional KQL joins\n- **Build detection rules** based on graph patterns you discover\n- **Scale analysis** by adjusting time windows and data volumes\n\n## Troubleshooting Tips\n\n- **No data found**: Some workspaces may not have SecurityEvent, SigninLogs, or SecurityAlert tables\n- **Authentication issues**: Try `az login` first, or check your service principal credentials\n- **Permission errors**: Ensure your account has Log Analytics Reader permissions\n- **Empty results**: Adjust time ranges - some workspaces have limited data retention\n\n## Resources\n\n- [Microsoft Sentinel KQL Reference](https://docs.microsoft.com/en-us/azure/data-explorer/kusto/query/)\n- [Graphistry Documentation](https://pygraphistry.readthedocs.io/)\n- [Azure Monitor Query Documentation](https://docs.microsoft.com/en-us/python/api/azure-monitor-query/)\n- [Sentinel Data Connectors](https://docs.microsoft.com/en-us/azure/sentinel/connect-data-sources)"


Next steps should include links, especially for most recommended

Try for free with your graphistry hub account and jupyter / google colab notebooks

Check out louie.ai for helping generating these queries and visualizations through genai

Check out the Kusto tutorial

Learn more about complementary pygraphistry features like UMAP for detecting patterns & outliers on events, GFQL for mining graphs once you've made them, and how to control graph colors and layouts

etc

lmeyerov · 2025-09-24T17:05:21Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+    "try:\n",
+    "    failed_logins = g.kql(failed_logins_query, timespan=timedelta(days=7))\n",
+    "    print(f\"Found {len(failed_logins)} users with multiple failed logins\")\n",
+    "    print(failed_logins.head())\n",


for all these df prints, you either want df.head() as the last cell line, or something like display(df.head()), so we get native notebook rendering

lmeyerov · 2025-09-24T17:06:07Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+    "        print(f\"Created graph with {len(nodes)} nodes and {len(edges)} edges\")\n",
+    "        \n",
+    "        # Plot the graph\n",
+    "        graph.plot()\n",


this needs to be the last line right? or a display() ?

lmeyerov · 2025-09-24T17:06:53Z

demos/demos_databases_apis/microsoft/sentinel/sentinel_security_analysis.ipynb

+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Graph Visualization\n",


make sure we have live plots in the output, so readthedocs will render

(unfortunately github won't)

lmeyerov · 2025-09-24T17:08:31Z

graphistry/plugins/kusto.py

        unwrap_nested: Optional[bool] = None,
-        single_table: bool = True
+        single_table: bool = True,
+        include_statistics: bool = False


i assume you tested kql didn't regress?

lmeyerov · 2025-09-24T17:10:50Z

graphistry/plugins/sentinel.py

+
+        dfs: List[pd.DataFrame] = []
+
+        for result in results:


can most of the wrangling in this file go away and be reused from kusto, so we're not maintaining 2 clones?

i'm not sure how close/different they are and how

ex: if just the initial connection changes, then we can parameterize kusto by the client connection

lmeyerov · 2025-09-24T17:14:03Z

graphistry/plugins/sentinel.py

+
+        return dfs
+
+    def kql_last(


i recommend removing kql_last() as conceptual bloat of surface area for users: seems unnecessary as can be folded into kql() as part of timespan param, like if timespan=... is an int, autoconvert it to hours

lmeyerov · 2025-09-24T17:15:50Z

graphistry/plugins/sentinel.py

+        # Use Usage table to get all table names - this avoids union conflicts
+        query = """
+        Usage
+        | where TimeGenerated > ago(30d)


TimeGenerated > ago(30d) is surprising to me

If desired as a convenience, add an optional param with a less surprising default? Eg, if not specified, drop that constraint?

lmeyerov · 2025-09-24T17:17:39Z

graphistry/plugins/sentinel.py

+        """
+        return self.kql(query, timespan=timedelta(days=30))
+
+    def sentinel_schema(self, table: str) -> pd.DataFrame:


helpful to document expected returned dtypes

lmeyerov · 2025-09-24T17:18:10Z

graphistry/plugins/sentinel.py

+                print(schema[['ColumnName', 'DataType']])
+        """
+        query = f"{table} | getschema"
+        return self.kql(query, timespan=timedelta(minutes=5))


why timespan here?

lmeyerov · 2025-09-24T17:19:12Z

graphistry/plugins/sentinel.py

+
+            # Check for partial failures
+            if response.status == LogsQueryStatus.PARTIAL:
+                logger.warning(f"Query returned partial results: {response.partial_error}")


should warnings raise too? or we make it an optional default-off param to raise on warnings on kql+here ?

i'm unsure how freq & fatal warnings etc are here

if a logger.warning, louie wouldn't see it...

lmeyerov · 2025-09-24T17:20:48Z

graphistry/plugins/sentinel.py

+            if response.status == LogsQueryStatus.PARTIAL:
+                logger.warning(f"Query returned partial results: {response.partial_error}")
+            elif response.status == LogsQueryStatus.FAILURE:
+                raise SentinelQueryError(f"Query failed: {response.partial_error}")


I wonder if we can propagate the partial result / response too so louie can inspect deeper.. esp in warning case..

lmeyerov · 2025-09-24T17:22:27Z

graphistry/plugins/sentinel.py

+    from azure.monitor.query import LogsQueryClient
+
+    try:
+        assert cfg.workspace_id is not None, "workspace_id is not set"


How does this get set? Give a more actionable error as may bubble up to user

lmeyerov · 2025-09-24T17:22:48Z

graphistry/plugins/sentinel.py

+
+        if cfg.credential:
+            credential = cfg.credential
+            logger.info("Using custom credential object for Sentinel")


debug not info

lmeyerov · 2025-09-24T17:22:58Z

graphistry/plugins/sentinel.py

+                client_id=cfg.client_id,
+                client_secret=cfg.client_secret
+            )
+            logger.info(f"Using Service Principal authentication for workspace {cfg.workspace_id}")


debug not info

lmeyerov · 2025-09-24T17:23:29Z

graphistry/plugins/sentinel.py

+            logger.info("You will be prompted to visit a URL and enter a code to authenticate")
+        else:
+            credential = DefaultAzureCredential()
+            logger.info(f"Using DefaultAzureCredential (Azure CLI, Managed Identity, etc.) for workspace {cfg.workspace_id}")


debug not info

lmeyerov · 2025-09-24T17:23:55Z

graphistry/plugins/sentinel.py

+                tenant_id=cfg.tenant_id  # Optional, uses common tenant if not provided
+            )
+            logger.info(f"Using Device Code authentication for workspace {cfg.workspace_id}")
+            logger.info("You will be prompted to visit a URL and enter a code to authenticate")


this goes to the user for action afaict, so info seems right, and/or maybe print, unsure

lmeyerov · 2025-09-24T17:24:41Z

graphistry/plugins/sentinel.py

+    return isinstance(val, (dict, list))
+
+
+def _unwrap_nested(result: SentinelQueryResult) -> pd.DataFrame:


yeah all this kind of stuff we want to combine as much as we can w/ kusto impl...

lmeyerov · 2025-09-24T17:25:37Z

graphistry/plotter.py

 #       to pass along args/kwargs to the next mixin in the chain
 class Plotter(
-    KustoMixin, SpannerMixin,
+    SentinelMixin, KustoMixin, SpannerMixin,


won't this override kusto as both have kql()?

lmeyerov · 2025-09-24T17:26:40Z

setup.py

    'nodexl': ['openpyxl==3.1.0', 'xlrd'],
    'jupyter': ['ipython'],
    'spanner': ['google-cloud-spanner'],
+    'sentinel': ['azure-monitor-query>=1.2.0', 'azure-identity>=1.12.0'],


are we sure about these restrictions or can be unpinned?

we prob want to preinstall pinned in graphistry and louie, so guidance here helps too

lmeyerov

See inline comments for generic things I noticed (use judgement if critical)

Most prominent I would recommend looking at:

does this break kusto, eg, overrides kusto() call, and thus a shared impl would make more sense
... there's a bunch of df wrangling, can that be shared with the kusto() impl?
Next Steps section in notebook seems untargeted, should be clickable links to more pyg/louie stuff (+ hub if a non-user)

SinsBre added 17 commits September 22, 2025 09:28

feat: Add Sentinel type definitions and data structures

acf451e

- Add SentinelConfig dataclass for connection configuration - Add SentinelQueryResult class for query results - Add custom exceptions for connection and query errors - Support for Azure authentication credentials

feat: Integrate SentinelMixin with Plotter class

31555fd

- Add SentinelMixin import to plotter.py - Include SentinelMixin in Plotter class hierarchy - Update documentation to list Sentinel integration

feat: Add Sentinel dependencies to setup.py

b32f4a4

- Add azure-monitor-query>=1.2.0 and azure-identity>=1.12.0 to sentinel extras - Add 'Sentinel' to package keywords for discoverability

feat: Update Graphistry authentication to use personal keys

4dfd959

- Replace username/password with personal_key_id/personal_key_secret - Update example.env with new credential format - Use modern Graphistry authentication method - Maintain security with environment variables

fix: Export Sentinel methods at module level

fe8ef5d

- Add configure_sentinel and sentinel_from_client to __init__.py - Add corresponding methods in pygraphistry.py - Fix module-level access to Sentinel functionality

fix: Add module-level function assignments for Sentinel

cb72c6f

- Add configure_sentinel and sentinel_from_client assignments - Follow same pattern as other plugins (Kusto, Spanner) - Enable direct import from graphistry module

fix: Update notebook to use correct DataType column for table listing

c6fe14f

- Fix references from TableName to DataType column - Clean up notebook code for consistent formatting - Add device authentication example - Ensure table listing and schema queries work correctly

fix: Use correct Graphistry encoding methods in notebook

39c368e

- Replace bind() with encode_point_color() and encode_edge_color() - Use encode_edge_size() for edge weight visualization - Fix method signatures to match Graphistry API

SinsBre requested review from lmeyerov and DataBoyTX September 23, 2025 08:59

SinsBre added 4 commits September 23, 2025 11:40

fix: Add missing SentinelMixin import in test file

e3925c2

- Import SentinelMixin to fix F821 undefined name errors - Resolves flake8 lint issues in test_sentinel.py

fix: Resolve all remaining lint issues

53e1f17

- Fix W504: Move binary operator before line break - Fix W292: Add newlines at end of all files - Fix F841: Remove unused local variables in tests - All flake8 lint issues now resolved

fix: Resolve mypy type checking errors for Sentinel plugin

c5a165d

fix: Resolve kql method signature conflict between SentinelMixin and …

763979b

…KustoMixin

lmeyerov reviewed Sep 24, 2025

View reviewed changes

lmeyerov requested changes Sep 24, 2025

View reviewed changes

		return isinstance(val, (dict, list))


		def _unwrap_nested(result: SentinelQueryResult) -> pd.DataFrame:

Feature/sentinel kql #743

Are you sure you want to change the base?

Feature/sentinel kql #743

Conversation

SinsBre commented Sep 23, 2025

Device code (interactive)

Service Principal (production)

DefaultAzureCredential (flexible)

Configure connection

Query failed logins

Create graph visualization

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lmeyerov Sep 24, 2025 •

edited

Loading

lmeyerov Sep 24, 2025 •

edited

Loading

lmeyerov Sep 24, 2025 •

edited

Loading

lmeyerov Sep 24, 2025 •

edited

Loading

lmeyerov Sep 24, 2025 •

edited

Loading

lmeyerov Sep 24, 2025 •

edited

Loading