Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

Python: Bug: Anthropic, Bedrock, Google AI and Vertex AI tool formatters ignore include_in_function_choices=False

Open Beginner friendly
  • Is something wrong?
#14,615 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 3 days

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
72/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
aws, gcp, python
Domain
ai, backend

Research direction

Start with the OpenAI formatter in semantic_kernel/connectors/ai/function_calling_utils.py, which already filters on include_in_function_choices, and copy that filter into the four formatters the issue names: anthropic/services/utils.py, bedrock/services/model_provider/utils.py, google/google_ai/services/utils.py and google/vertex_ai/services/utils.py. Filter both properties and required. Done when the reproduction script in the issue shows only query for every provider and the new tests fail on main and pass with the change.

Written by the indexing model from the issue text.

Description

python triage

Describe the bug

A parameter marked with {"include_in_function_choices": False} is hidden from the model by the OpenAI formatters. The Anthropic, Bedrock, Google AI and Vertex AI formatters ignore the flag. They still send the parameter in the tool schema.

The decorator docs say a parameter with this flag "will not be included in the function choices". In function_calling_utils.py both OpenAI formatters filter on include_in_function_choices. The four other formatters read metadata.parameters without the filter.

Files on current main:

  • python/semantic_kernel/connectors/ai/anthropic/services/utils.py
  • python/semantic_kernel/connectors/ai/bedrock/services/model_provider/utils.py
  • python/semantic_kernel/connectors/ai/google/google_ai/services/utils.py
  • python/semantic_kernel/connectors/ai/google/vertex_ai/services/utils.py

To Reproduce

I ran this on microsoft/semantic-kernel main at 7a92cd4e7 (Python 3.13).

from typing import Annotated
from semantic_kernel import Kernel
from semantic_kernel.functions import kernel_function
from semantic_kernel.connectors.ai.function_calling_utils import kernel_function_metadata_to_function_call_format as oa
from semantic_kernel.connectors.ai.anthropic.services.utils import kernel_function_metadata_to_function_call_format as an
from semantic_kernel.connectors.ai.google.google_ai.services.utils import kernel_function_metadata_to_google_ai_function_call_format as ga
from semantic_kernel.connectors.ai.google.vertex_ai.services.utils import kernel_function_metadata_to_vertex_ai_function_call_format as va
from semantic_kernel.connectors.ai.bedrock.services.model_provider.utils import update_settings_from_function_choice_configuration as bd
from semantic_kernel.connectors.ai.bedrock.bedrock_prompt_execution_settings import BedrockChatPromptExecutionSettings
from semantic_kernel.connectors.ai.function_choice_type import FunctionChoiceType
from semantic_kernel.connectors.ai.function_call_choice_configuration import FunctionCallChoiceConfiguration


class P:
    @kernel_function(name="lookup", description="Look up a user")
    def lookup(
        self,
        query: Annotated[str, "What to look up"],
        token: Annotated[str, "Internal token", {"include_in_function_choices": False}] = "",
    ) -> str:
        return query


meta = Kernel().add_plugin(P(), "p").functions["lookup"].metadata
print("OpenAI    :", list(oa(meta)["function"]["parameters"]["properties"]))
print("Anthropic :", list(an(meta)["input_schema"]["properties"]))
print("GoogleAI  :", list(ga(meta)["parameters"]["properties"]))
print("VertexAI  :", list(type(va(meta)).to_dict(va(meta))["parameters"]["properties"]))
s = BedrockChatPromptExecutionSettings()
bd(FunctionCallChoiceConfiguration(available_functions=[meta]), s, FunctionChoiceType.AUTO)
print("Bedrock   :", list(s.tools[0]["toolSpec"]["inputSchema"]["json"]["properties"]))

Output:

OpenAI    : ['query']
Anthropic : ['query', 'token']
GoogleAI  : ['query', 'token']
VertexAI  : ['token', 'query']
Bedrock   : ['query', 'token']

Expected behavior

All formatters hide token, as the OpenAI one does. The model should only see query.

Impact

The parameter name and description are sent to the model on these providers. The model can then try to fill a value that the caller meant to supply itself. It also adds tokens to every request.

Fix

The fix is small. Filter on include_in_function_choices in the four formatters, for both properties and required. I have a branch with tests that fail on main and pass with the change: https://github.com/VANDRANKI/semantic-kernel/tree/fix/non-openai-formatters-include-in-function-choices

I can open a PR from this branch if the maintainers want it.

Dominant language
C#
Stars
28.6k
Forks
4.8k
Avg merge
2d 12h
Merged PRs (30d)
22

Getting set up

Open in Codespaces

Starts the project's dev container in your browser, under your own GitHub account.

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from microsoft/semantic-kernel

All issues in microsoft/semantic-kernel

Similar issues

More C# issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.