Skip to content

.Net: New Feature: Gemini Grounding with Google Search #11735

New issue
New issue
Open
Feature
#11804
Open
.Net: New Feature: Gemini Grounding with Google Search#11735
Feature
#11804
Assignees
rogerbarreto
Labels
.NETIssue or Pull requests regarding .NET codeai connectorAnything related to AI connectors
@scherervictor

Description

@scherervictor
scherervictor
opened on Apr 25, 2025

name: Gemini - Grouding with Google Search
about: Gemini supports grouding with Google Search if configured properly on the request, would be nice to have the option to configure this param (Doc reference)

Activity

scherervictor
added the
Feature
issue type on Apr 25, 2025
markwallace-microsoft
added
.NETIssue or Pull requests regarding .NET code
triage
on Apr 25, 2025
markwallace-microsoft
assigned
rogerbarreto
on Apr 25, 2025
markwallace-microsoft
added
ai connectorAnything related to AI connectors
and removed
triage
on Apr 25, 2025
markwallace-microsoft

markwallace-microsoft commented on Apr 25, 2025

@markwallace-microsoft
markwallace-microsoft
on Apr 25, 2025
Member

@scherervictor Would you be interested in contributing this functionality to the Semantic Kernel?

scherervictor

scherervictor commented on Apr 25, 2025

@scherervictor
scherervictor
on Apr 25, 2025
Author

@markwallace-microsoft sure! I could try to do that.

dmm-l-mediehus

dmm-l-mediehus commented on Apr 29, 2025

@dmm-l-mediehus
dmm-l-mediehus
on Apr 29, 2025

@markwallace-microsoft If the docs is:

curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=$GOOGLE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
      "contents": [
          {
              "parts": [
                  {"text": "What is the current Google stock price?"}
              ]
          }
      ],
      "tools": [
          {
              "google_search": {}
          }
      ]
  }' > result.json

cat result.json

Would it be equivelent to this in SK?

GeminiPromptExecutionSettings settings = new()
{
    ModelId = AIModel.Google_Gemini25ProPreview.ToModelIdentifier(),
    ServiceId = AIModel.Google_Gemini25ProPreview.ToModelIdentifier(),
    MaxTokens = 32768,
    ExtensionData = new Dictionary<string, object>()
    {
        ["tools"] = new List<Dictionary<string, object>>
        {
            new() 
            {
                ["google_search"] = new Dictionary<string, object>()
            }
        }
    }
};
scherervictor
linked a pull request that will close this issue on Apr 29, 2025
scherervictor

scherervictor commented on Apr 29, 2025

@scherervictor
scherervictor
on Apr 29, 2025 · edited by scherervictor
Author

@markwallace-microsoft @rogerbarreto I've created a PR implementing this, could you review it, please?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

.NETIssue or Pull requests regarding .NET codeai connectorAnything related to AI connectors

Type

Feature

Projects

No projects

Milestone

No milestone

Relationships

None yet

    Development

    Participants

    @rogerbarreto@scherervictor@dmm-l-mediehus@markwallace-microsoft

    Issue actions
      .Net: New Feature: Gemini Grounding with Google Search · Issue #11735 · microsoft/semantic-kernel