> For the complete documentation index, see [llms.txt](https://docs.devarmor.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.devarmor.com/modules/code-reviewer/ai-guardrails.md).

# AI Guardrails

Turn security policies into rules your AI coding assistant follows.

AI Guardrails generates rule files for AI coding assistants — like Cursor and Augment Code — directly from your security policies, so the same requirements Code Review enforces on a pull request are available to an AI assistant while it's writing the code in the first place.

## Generating rules

You can generate a rule file:

* From an overall security policy, covering everything the policy applies to.
* From a specific design review, scoped to just that feature.

When setting up a rule, you can browse the repository's files and folders to scope it precisely.

## Scoping

Rules can be scoped to a specific **threat model**, **threat model version**, or **design review**, and filtered accordingly — so a rule only surfaces where it's actually relevant, instead of applying blanket guidance across every repository.

## From rule to pull request

You can open a pull request directly from a guardrail rule run, without leaving DevArmor, to apply the rule's suggested change.