> For the complete documentation index, see [llms.txt](https://help.nightfall.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.nightfall.ai/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-policy-configurations/ai-agent-and-inference-hooks-policies.md).

# AI agent and Inference Hooks policies

Which AI Agent Security and Inference Hooks policies to pair with each endpoint recommendation.

Endpoint triggers see browser uploads, paste, desktop apps, cloud sync, git push, removable media, and print. They do not see what a coding agent does inside Claude Code, Cursor, or VS Code Copilot: its prompts, MCP tool calls and responses, terminal commands, and file writes. They also do not see Claude Enterprise on devices without the Nightfall agent.

AI Agent Security policies cover that gap. The full templates are on [Hook policy recommendations](/data-exfiltration-prevention/ai-agent-security/ai-governance/hook-policy-recommendations.md). This page shows which ones to add next to the endpoint policies you already have.

### Pair each endpoint policy with its agent policy

| If you have                                      | Add                                                               | Why                                                                                                                                                                                  |
| ------------------------------------------------ | ----------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Core policies 1 and 4 (unsanctioned apps and AI) | Agent policy 3, Inference Hooks, if you use Claude Enterprise     | If Claude is in your Sanctioned Apps collection, policies 1 and 4 skip it. Inference Hooks control what goes into it, on every device and in the desktop and mobile apps             |
| Recommendation 1 (credentials)                   | Agent policies 1 and 3                                            | Policy 1 stops a secret before a coding agent sends it, passes it to a tool, reads it back from a tool, or runs it in a command. Policy 3 stops it before Claude reads it            |
| Recommendation 2 (source code)                   | Agent policy 5, the MCP Gateway                                   | Coding agents read and write code by design, so a source code rule on them would block normal work. Control where code can go instead: which remote tools exist and what they can do |
| Recommendation 3 (PHI and PCI)                   | Agent policies 2, 3, and 4                                        | Customer records reach agents through pasted prompts and MCP tools that query databases and tickets                                                                                  |
| Recommendation 4 (confidential documents)        | Agent policy 3, with file classifiers added after the shadow week | Contracts and financial reports are uploaded to Claude as attachments. Inference Hooks scan the text extracted from them                                                             |

### Three things that differ from endpoint policies

* **One action per policy.** Block or Monitor applies to every hook type in the policy. Model Responses can only record, so keep it in a separate Monitor policy.
* **No end-user notification email and no override with justification.** The person sees the policy's custom block message in their client. Write it so they know what to do next.
* **Content scanning off is enforced, as on other endpoint policies.** Block denies every event the policy covers, and Monitor records every event. Turn content scanning on and add detection rules.

Create agent policies under **Configuration > Policies > Exfiltration**, with **AI Agent Security** as the integration. Setup is on [Setup and Installation](/data-exfiltration-prevention/ai-agent-security/ai-governance/auditability-and-control/setup-and-installation.md) for device hooks and [Connect Inference Hooks](/data-exfiltration-prevention/ai-agent-security/ai-governance/auditability-and-control/setup-and-installation/connect-inference-hooks.md) for Claude Enterprise.


---

# 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://help.nightfall.ai/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-policy-configurations/ai-agent-and-inference-hooks-policies.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.