Python: Bug: ConversationSummaryPlugin Python

[haithamshahin333]

Describe the bug
There is not a clear way to leverage this plugin given the function arguments expect both the Kernel and KernelArguments passed to it.

To Reproduce

    arguments["input"] = 'this is a story to be summarized...'
    
    await kernel.invoke(plugin_name="summarizer", function_name="SummarizeConversation", arguments=arguments)

Expected behavior
A summary to be produced

Screenshots
Consistent error on function calling debug output stating that 3 arguments are expected, but only one is provided.

Platform

• OS: [e.g. Windows, Mac]
• IDE: [e.g. Visual Studio, VS Code]
• Language: [e.g. C#, Python]
• Source: [e.g. NuGet package version 0.1.0, pip package version 0.1.0, main branch of repository]

Additional context
Is there a code example of how to call this plugin?

[sphenry]

@eavanvalkenburg can you take a look?

[eavanvalkenburg]

@haithamshahin333 could you share a bit more of the broader script you are using for this, we haven't spend much time on this one, but the tests do work for it. The only thing I see different in your small snippet is that we explicitly do: arguments = KernelArguments(input=CHAT_TRANSCRIPT) but you might have created the KernelArguments earlier. This test works with OpenAI models: https://github.com/microsoft/semantic-kernel/blob/main/python/tests/integration/completions/test_conversation_summary_plugin.py

[moonbox3]

Closing the issue as we've provided some sample code on how to structure a call to the plugin. Please re-open if you continue to have issues, or file a new issue. Thank you.

[haithamshahin333]

Hi Team,

Apologies for the delay on getting back to you - to provide some clarity @eavanvalkenburg I'm attempting to invoke the plugin with an agent through auto invocation. Given that the function is annotated with both the Kernel and KernelArguments as parameters, it is failing

If I manually invoke the function through kernel.invoke I don't need to pass the Kernel and all works well