diff --git a/docs/workflow-failure-analysis.md b/docs/workflow-failure-analysis.md
new file mode 100644
index 00000000..3a896b4b
--- /dev/null
+++ b/docs/workflow-failure-analysis.md
@@ -0,0 +1,80 @@
+# SDK Consistency Review Agent - Workflow Failure Analysis
+
+**Workflow Run:** [#21486968231](https://github.com/github/copilot-sdk/actions/runs/21486968231)  
+**Date:** January 29, 2026  
+**Status:** Failed (Race Condition)
+
+## Executive Summary
+
+The SDK Consistency Review Agent workflow failed due to a race condition in the gh-aw infrastructure where the PR branch was deleted (after merge) before the workflow could checkout the code. **However, this failure prevented the workflow from catching a real documentation inconsistency** - the .NET SDK was missing important Azure provider configuration documentation that was added to the other three SDKs.
+
+## What Happened
+
+### Timeline
+1. **16:49:13** - PR #260 opened (provider-info branch): "Update docs to reflect you need version for Azure Foundry"
+2. **16:50:46** - SDK Consistency Review workflow triggered by PR event
+3. **16:50:53** - **PR #260 merged** (branch deleted)
+4. **16:51:08** - Workflow step "Checkout PR branch" failed: `fatal: couldn't find remote ref provider-info`
+
+### The Failure
+```
+fatal: couldn't find remote ref provider-info
+##[error]Failed to checkout PR branch: The process '/usr/bin/git' failed with exit code 128
+```
+
+The workflow's checkout step tried to `git fetch origin provider-info` but the branch no longer existed because it was deleted as part of the merge process.
+
+## Root Cause Analysis
+
+### Infrastructure Issue
+The gh-aw workflow engine's `checkout_pr_branch.cjs` script uses the branch name for checkout instead of the commit SHA. This creates a race condition:
+
+- **Branch names** are ephemeral - they get deleted after PR merge
+- **Commit SHAs** persist indefinitely - they remain accessible even after branch deletion
+
+**Solution:** The gh-aw infrastructure should use `github.event.pull_request.head.sha` for checkout, which is always available and never gets deleted. This has been logged as a separate work item for the gh-aw team.
+
+### What the Workflow Missed
+
+PR #260 modified README documentation for three SDKs:
+- ✅ **Node.js SDK** - Added Azure provider configuration notes
+- ✅ **Python SDK** - Added Azure provider configuration notes  
+- ✅ **Go SDK** - Added Azure provider configuration notes
+- ❌ **.NET SDK** - **Not updated** (inconsistency)
+
+The changes included critical information:
+- Azure OpenAI endpoints must use `type: "azure"`, not `type: "openai"`
+- BaseURL should be just the host, not include `/openai/v1` path
+- Model parameter is required when using custom providers
+- Examples for Ollama, custom OpenAI-compatible APIs, and Azure OpenAI
+
+This is **exactly the kind of cross-SDK inconsistency** the workflow is designed to catch.
+
+## Resolution
+
+### Immediate Fix
+✅ **Added comprehensive Azure provider documentation to .NET SDK README** (commit: 9778e86)
+
+The .NET SDK now has:
+- Complete ProviderConfig field documentation
+- Ollama example with local provider usage
+- Custom OpenAI-compatible API example
+- Azure OpenAI example with important notes
+- All the same warnings and best practices as other SDKs
+
+### Future Prevention
+📋 **Proposed work item:** Fix gh-aw race condition by using commit SHA instead of branch name for checkout
+
+This will prevent future workflow failures when PRs are merged quickly, ensuring reviews can complete even for fast-moving PRs.
+
+## Key Takeaways
+
+1. **The workflow failure was a symptom, not the disease** - The real issue was the documentation gap across SDKs
+2. **Fast merges expose infrastructure limitations** - The gh-aw checkout logic needs to be more resilient
+3. **Consistency reviews are valuable** - Even though the workflow failed, manual analysis found the exact kind of issue it was designed to catch
+
+## Recommendations
+
+1. **For gh-aw team:** Implement SHA-based checkout with fallback to branch name
+2. **For SDK maintainers:** When updating documentation across SDKs, always check all four implementations (Node.js, Python, Go, .NET)
+3. **For this repository:** Consider adding a docs linting check that validates critical sections exist in all SDK READMEs
diff --git a/dotnet/README.md b/dotnet/README.md
index c63bb120..c0ca3e71 100644
--- a/dotnet/README.md
+++ b/dotnet/README.md
@@ -430,20 +430,73 @@ await session.SendAsync(new MessageOptions
 
 ### Bring Your Own Key (BYOK)
 
-Use a custom API provider:
+The SDK supports custom OpenAI-compatible API providers (BYOK - Bring Your Own Key), including local providers like Ollama. When using a custom provider, you must specify the `Model` explicitly.
+
+**ProviderConfig fields:**
+
+- `Type` (string): Provider type - `"openai"`, `"azure"`, or `"anthropic"` (default: `"openai"`)
+- `BaseUrl` (string): API endpoint URL (required)
+- `ApiKey` (string): API key (optional for local providers like Ollama)
+- `BearerToken` (string): Bearer token for authentication (takes precedence over `ApiKey`)
+- `WireApi` (string): API format for OpenAI/Azure - `"completions"` or `"responses"` (default: `"completions"`)
+- `Azure` (AzureProviderOptions): Azure-specific options with `ApiVersion` (default: `"2024-10-21"`)
+
+**Example with Ollama:**
+
+```csharp
+var session = await client.CreateSessionAsync(new SessionConfig
+{
+    Model = "deepseek-coder-v2:16b",  // Required when using custom provider
+    Provider = new ProviderConfig
+    {
+        Type = "openai",
+        BaseUrl = "http://localhost:11434/v1",  // Ollama endpoint
+        // ApiKey not required for Ollama
+    }
+});
+
+await session.SendAsync(new MessageOptions { Prompt = "Hello!" });
+```
+
+**Example with custom OpenAI-compatible API:**
 
 ```csharp
 var session = await client.CreateSessionAsync(new SessionConfig
 {
+    Model = "gpt-4",
     Provider = new ProviderConfig
     {
         Type = "openai",
-        BaseUrl = "https://api.openai.com/v1",
-        ApiKey = "your-api-key"
+        BaseUrl = "https://my-api.example.com/v1",
+        ApiKey = Environment.GetEnvironmentVariable("MY_API_KEY")
+    }
+});
+```
+
+**Example with Azure OpenAI:**
+
+```csharp
+var session = await client.CreateSessionAsync(new SessionConfig
+{
+    Model = "gpt-4",
+    Provider = new ProviderConfig
+    {
+        Type = "azure",  // Must be "azure" for Azure endpoints, NOT "openai"
+        BaseUrl = "https://my-resource.openai.azure.com",  // Just the host, no path
+        ApiKey = Environment.GetEnvironmentVariable("AZURE_OPENAI_KEY"),
+        Azure = new AzureProviderOptions
+        {
+            ApiVersion = "2024-10-21"
+        }
     }
 });
 ```
 
+> **Important notes:**
+> - When using a custom provider, the `Model` parameter is **required**. The SDK will throw an error if no model is specified.
+> - For Azure OpenAI endpoints (`*.openai.azure.com`), you **must** use `Type = "azure"`, not `Type = "openai"`.
+> - The `BaseUrl` should be just the host (e.g., `https://my-resource.openai.azure.com`). Do **not** include `/openai/v1` in the URL - the SDK handles path construction automatically.
+
 ## Error Handling
 
 ```csharp