@@ -27,7 +27,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - **tests**: Expand DevUI coverage for reasoning chunks without text ([#9543](https://github.com/microsoft/agent-framework/pull/9543))
 
 ### Fixed
-- **agent-framework-core**: Rebuild nested content, accept `Literal` workflow-edge sources, close abandoned inner streams, pass stream errors into nested streams during closure, make Memory and Todo writes atomic, bind background completion waits to execution identity, support keyword-only functional-workflow contexts, isolate MCP task contexts, normalize tool-name whitespace, improve settled approval matching, recover owned MCP connections safely, and omit span-only tool attributes from invocation metrics ([#9304](https://github.com/microsoft/agent-framework/pull/9304), [#9355](https://github.com/microsoft/agent-framework/pull/9355), [#9331](https://github.com/microsoft/agent-framework/pull/9331), [#9397](https://github.com/microsoft/agent-framework/pull/9397), [#9412](https://github.com/microsoft/agent-framework/pull/9412), [#9442](https://github.com/microsoft/agent-framework/pull/9442), [#9451](https://github.com/microsoft/agent-framework/pull/9451), [#9476](https://github.com/microsoft/agent-framework/pull/9476), [#9557](https://github.com/microsoft/agent-framework/pull/9557), [#9535](https://github.com/microsoft/agent-framework/pull/9535), [#9583](https://github.com/microsoft/agent-framework/pull/9583), [#9592](https://github.com/microsoft/agent-framework/pull/9592))
+- **agent-framework-core**: Rebuild nested content, accept `Literal` workflow-edge sources, close abandoned inner streams, pass stream errors into nested streams during closure, make Memory and Todo writes atomic, bind background completion waits to execution identity, support keyword-only functional-workflow contexts, isolate MCP task contexts, normalize tool-name whitespace, recommend `include_detailed_errors=True` in function-execution warnings only when detailed errors are disabled, improve settled approval matching, recover owned MCP connections safely, and omit span-only tool attributes from invocation metrics ([#9304](https://github.com/microsoft/agent-framework/pull/9304), [#9355](https://github.com/microsoft/agent-framework/pull/9355), [#9331](https://github.com/microsoft/agent-framework/pull/9331), [#9397](https://github.com/microsoft/agent-framework/pull/9397), [#9412](https://github.com/microsoft/agent-framework/pull/9412), [#9442](https://github.com/microsoft/agent-framework/pull/9442), [#9451](https://github.com/microsoft/agent-framework/pull/9451), [#9472](https://github.com/microsoft/agent-framework/pull/9472), [#9476](https://github.com/microsoft/agent-framework/pull/9476), [#9557](https://github.com/microsoft/agent-framework/pull/9557), [#9535](https://github.com/microsoft/agent-framework/pull/9535), [#9583](https://github.com/microsoft/agent-framework/pull/9583), [#9592](https://github.com/microsoft/agent-framework/pull/9592))
 - **agent-framework-ag-ui**: Preserve every queued approval result and repeated single-chunk text deltas ([#8851](https://github.com/microsoft/agent-framework/pull/8851), [#9339](https://github.com/microsoft/agent-framework/pull/9339))
 - **agent-framework-anthropic**: Keep bash tool-result files nested inside their shell tool result ([#9343](https://github.com/microsoft/agent-framework/pull/9343))
 - **agent-framework-azure-cosmos-memory**: Reject non-positive result limits and confidence values outside the supported range ([#9318](https://github.com/microsoft/agent-framework/pull/9318))