AI Agent Endpoint Credentials
AI Agent Endpoint credentials store the application programming interface (API) key that authenticates Cisco Workflows with a large language model (LLM) provider. Create the credential as an account key, then assign it to an AI Agent Endpoint target.
Prerequisite
Create an API key with the LLM provider before adding the credential. Each provider issues the key from its own console, and the key grants access billed to the provider account.
Supported credential types
The AI Agent Endpoint target supports one credential type per provider. The following table maps each credential type to the location where the key is obtained.
|
Credential type |
Provider |
Where to obtain the key |
|
OpenAI API Key |
OpenAI |
|
|
Anthropic API Key |
Anthropic |
|
|
Google Gemini API Key |
Google Gemini |
|
|
Azure OpenAI API Key |
Azure OpenAI |
Azure Portal, under the OpenAI resource |
|
OpenRouter API Key |
OpenRouter |
AWS Bedrock uses AWS Credentials rather than a single API key. Create AWS Credentials when the target provider is AWS Bedrock.
Procedure
Perform the following steps to create an AI provider credential:
-
From the left navigation pane, click Automation > Targets.
-
In the Account Key section, click New account key.
-
From the Account Key Type drop-down list, choose the credential type that matches the provider, such as OpenAI API Key.
-
Enter a unique display name and a description for the account key.
-
In the API Key field, paste the API key obtained from the provider. The field hides the value after entry.
-
Click Save to save the account key.
Next steps
- Assign the account key to an AI Agent Endpoint target. See AI Agent Endpoint Target.
- Rotate the API key in the provider console and update the account key when the key changes.