semantic-kernel
semantic-kernel copied to clipboard
microsoft
Integrate cutting-edge LLM technology quickly and easily into your apps
We are looking to pass in user information for our Chat Completion calls, is there any documentation on how to pass in additional parameters like this to our Azure Open...
ITextSearch accepts TextSearchOptions, which currently has an optional TextSearchFilter, which relies on the legacy non-LINQ vector search filtering feature. We need to modify it to use the new search interface....
--- name: Agent Memory; AIContextProvider implementation for Python about: Update the SK Python SDK to add the C# equivalent of AIContextProviders --- Related to #10562 that is stale and #10100...
--- name: Magentic orchestration to take a MagenticContext as an input --- Currently, the Magentic orchestration doesn't allow users to pass a chat history to invoke the orchestration. This limits...
The call to LLM comes back with empty message at sometimes. It's not making any tool calls or giving any response - comes back back with empty response. this works...
…ync support ### Motivation and Context This change is required to enable Semantic Kernel users to store and retrieve embeddings using Oracle databases. Currently, Semantic Kernel supports vector storage for...
**Describe the bug** The schema sent to the model for the tools when adding a [PromptTemplateConfig ](https://github.com/microsoft/semantic-kernel/blob/7d72f890f6598183ecbefbe31fdb2fb73e2a8775/python/semantic_kernel/prompt_template/prompt_template_config.py#L22)as a KernelFunction is incorrect. [PromptTemplateConfig.get_kernel_parameter_metadata](https://github.com/microsoft/semantic-kernel/blob/7d72f890f6598183ecbefbe31fdb2fb73e2a8775/python/semantic_kernel/prompt_template/prompt_template_config.py#L79) creates a KernelParameterMetadata, setting [type_ = InputVariable.json_schema](https://github.com/microsoft/semantic-kernel/blob/7d72f890f6598183ecbefbe31fdb2fb73e2a8775/python/semantic_kernel/prompt_template/prompt_template_config.py#L86). Then,...
**Describe the bug** When invoking a single AzureAIAgent with an MCP-tool calling capability via Semantic Kernel (SK) for multiple sequential function calls, the agent hallucinates and produces inaccurate function arguments...
**Describe the bug** When you instantiate a ChatCompletionAgent with KernelArguments (for example to enforce a Pydantic response schema), the registered MCP plugin is completely ignored. No function calls are emitted...
**Describe the bug** In projects with `enable` when using `AllowStrictSchemaAdherence` the schema does not add "null" for nullable parameters. For example a plugin function : `void MyFunction(string? something)` will generate...
