Google Gemini
Secuvon calls Google Gemini at its fixed hosted endpoint with your own key, so the provider bills your account for every test it answers.
Setup
- 1
Create a Google Gemini key
Create an API key at aistudio.google.com (opens in a new tab) for an account with credit, and keep it to hand.
- 2
Save it in Secuvon
In Secuvon open Settings, then Credentials, choose Add credential, pick Google Gemini, name the credential and paste the key. It is encrypted at rest and never shown again.
- 3
Add the agent
Open Agents, choose Add agent, pick Google Gemini, choose its model and attach the saved credential.
- 4
Run an assessment
Open New assessment, choose the agent and start a profile. Choose a versioned profile, from Smoke 25 through the full 4,075-test corpus, within your contract capacity.
- 5
Read the report
The report lists the graded findings with the response evidence behind each one, and keeps tests that errored apart from the score.
Check the key first
This asks Google Gemini for the models your key can call. An error here will be an error in every test too.
curl -s "https://generativelanguage.googleapis.com/v1beta/models?key=$GEMINI_API_KEY"Secuvon calls the Gemini API with a Google AI Studio key, not a Vertex AI service account.
Common issues
The provider refuses the key
The tests that reached the provider error, the report says why, and none of them counts as a pass or a fail. Save a new key in Settings, then Credentials, and run the assessment again.
The provider rate-limits or runs out of credit mid-run
Tests the provider did not answer error and stay out of the score. If every test errored, the scan is Not scored rather than graded as zero.
The model id is wrong
Use a model id your provider account can call, exactly as the provider lists it. The key check above prints the ids your key can use.
Ready to assess a Google Gemini agent?
The API route for each step is in the documentation.
Request demo