> For the complete documentation index, see [llms.txt](https://docs.groundcover.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.groundcover.com/automate/agent-mode/requirements.md).

# Requirements & Compatibility

Deployment types and prerequisites for using Agent Mode

Agent Mode requires network access to a supported LLM endpoint and the authentication method that endpoint expects. The endpoint can be a native cloud LLM service, the Anthropic API, or an Anthropic-compatible proxy. This page covers which deployment types support Agent Mode and what setup is needed. See [Privacy & Security](https://docs.groundcover.com/automate/agent-mode/privacy-and-security.md#llm-provider) for provider-specific data handling.

## Supported Deployments

| Deployment Type                 | Agent Mode | Setup Required                                                                                                                                                         |
| ------------------------------- | ---------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **groundcover Cloud**           | Yes        | None                                                                                                                                                                   |
| **BYOC AWS**                    | Yes        | None. Model access is enabled automatically during validation when your account permits it ([Marketplace subscriptions](#aws-byoc-bedrock-marketplace-subscriptions)). |
| **BYOC GCP**                    | Yes        | [Enable models in Vertex AI](https://docs.groundcover.com/collect-data/byoc/setup-byoc-with-gcp/enable-ai-models-for-agent-mode.md)                                    |
| **BYOC Azure**                  | Yes        | [Deploy additional models in Foundry](https://docs.groundcover.com/collect-data/byoc/setup-byoc-with-azure/enable-ai-models-for-agent-mode.md)                         |
| **On-premises (AWS/GCP/Azure)** | Yes        | Use your own Anthropic API key or compatible endpoint, or contact your account team for a native cloud provider                                                        |
| **On-premises (other)**         | Yes        | Use your own Anthropic API key or compatible endpoint                                                                                                                  |

{% hint style="info" %}
Agent Mode is enabled by default on groundcover Cloud and BYOC deployments once provisioning completes. Model access, publisher terms, and charges depend on the models and cloud deployment types you choose. Apart from model validation, which may incur minimal inference charges, there are no inference costs until you use the Agent.
{% endhint %}

For on-premises deployments on AWS, GCP, or Azure, groundcover can provision the native LLM service: AWS Bedrock, Google Cloud Vertex AI, or Microsoft Foundry. You can also use a direct Anthropic API key or a custom Anthropic-compatible endpoint on any supported backend. All configurations require network access to the selected LLM provider or proxy.

## Custom LLM providers

You can override the native cloud provider for an individual backend from **Settings > AI & Agents > Management**.

### Anthropic API key

Agent Mode supports using your own Anthropic API key for inference on any backend where AI features are available. This is useful when you want usage billed through your Anthropic account or when a native cloud-provider LLM service is not available for your deployment.

Workspace administrators configure the key separately for each backend. Inference requests are sent directly to Anthropic after the key is configured. See [Configuring Settings](https://docs.groundcover.com/automate/agent-mode/configuring-settings.md#anthropic-api-key) for the self-service setup and key rotation steps.

### Custom Anthropic domain

**Custom Anthropic Domain** sends inference requests to an Anthropic-compatible endpoint using its base URL and API key. You can use LiteLLM, Bifrost, or any other proxy or gateway that implements the Anthropic Messages API and exposes the Claude models Agent Mode uses.

Your groundcover deployment must have network access to the custom endpoint. The endpoint must support streaming Anthropic Messages API responses. See [Configuring Settings](https://docs.groundcover.com/automate/agent-mode/configuring-settings.md#choose-an-llm-provider) for setup instructions and endpoint examples.

## GCP BYOC setup

GCP BYOC deployments require access to the selected models in your project's configured Vertex AI region. Complete any required model enablement and permissions before using them. See [Enable AI models for Agent Mode](https://docs.groundcover.com/collect-data/byoc/setup-byoc-with-gcp/enable-ai-models-for-agent-mode.md) for step-by-step instructions.

## AWS BYOC: Bedrock Marketplace subscriptions

AWS Bedrock serves the models Agent Mode uses through AWS Marketplace. The first time Agent Mode invokes a model in your BYOC account, Bedrock subscribes the account to the model's Marketplace listing. Each model gets its own subscription, so you see one for every Claude model Agent Mode uses, including models users pick in the model picker.

AWS Marketplace and AWS CloudTrail show these subscriptions as created by the agent service IAM role, `groundcover-agent-service-<install-id>-sa`. groundcover provisions the role with the following Marketplace permissions:

```json
[
  {
    "Sid": "marketplaceSubscribeUnsubscribe",
    "Effect": "Allow",
    "Action": ["aws-marketplace:Subscribe", "aws-marketplace:Unsubscribe"],
    "Resource": "*",
    "Condition": { "StringEquals": { "aws:CalledViaLast": "bedrock.amazonaws.com" } }
  },
  {
    "Sid": "marketplaceViewSubscriptions",
    "Effect": "Allow",
    "Action": ["aws-marketplace:ViewSubscriptions"],
    "Resource": "*",
    "Condition": { "StringEquals": { "aws:CalledViaLast": "bedrock.amazonaws.com" } }
  }
]
```

The `aws:CalledViaLast` condition limits the role to subscriptions that Bedrock makes on its behalf while invoking a model. The role cannot subscribe to Marketplace products outside Bedrock. A subscription has no charge of its own: Bedrock bills inference for the tokens Agent Mode uses.

If a service control policy or a private marketplace in your AWS organization blocks these subscriptions, Agent Mode cannot invoke the affected models. Allow the foundation model provider's Bedrock listings for the BYOC account.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.groundcover.com/automate/agent-mode/requirements.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
