RAMPART

About

RAMPART is ranked #14 of 29 in AI security testing tools on Inferse. It runs on API, Linux, macOS, Windows.

Compared on AI security testing tools

Prompt injection tests

Yesmicrosoft.github.io

Data leakage tests

Yesmicrosoft.github.io

Unsafe output tests

Yesmicrosoft.github.io

Custom test cases

Yesmicrosoft.github.io

Deployment mode

self_hostedmicrosoft.github.io

Facts

Purpose

RAMPART is a pytest-native safety testing framework for agentic AI applications that orchestrates attacks or probes, evaluates outcomes, and reports results.microsoft.github.io · 8 Oct 2026

Attack and probe tests

Its two test categories are attacks, where a detected attack means UNSAFE, and probes, where detected expected behavior means SAFE.microsoft.github.io · 8 Oct 2026

Included tests

The documented built-in attack is cross-prompt injection (XPIA), and the documented probe verifies expected agent behavior.microsoft.github.io · 8 Oct 2026

Integration model

Users provide an adapter connecting their agent to RAMPART, which supplies execution strategies, evaluators, pytest integration, and reporting.microsoft.github.io · 8 Oct 2026

PyRIT

RAMPART uses PyRIT for LLM interaction, prompt normalization, and conversation memory, and installs PyRIT as a dependency.microsoft.github.io · 8 Oct 2026

OneDrive

An optional onedrive extra installs dependencies for using the built-in OneDriveSurface to place XPIA payloads in OneDrive.microsoft.github.io · 8 Oct 2026

Python requirement

Installation requires Python 3.11 or later and uses uv or pip for package management.microsoft.github.io · 8 Oct 2026

Supported installation systems

The installation instructions provide commands for Linux, macOS, and Windows.microsoft.github.io · 8 Oct 2026

pytest plugin

RAMPART registers automatically as a pytest plugin through the pytest11 entry point, without requiring conftest.py activation configuration.microsoft.github.io · 8 Oct 2026

Reporting

RAMPART provides structured JSON reporting, including a built-in JSON file report sink and a hook for custom sinks.microsoft.github.io · 8 Oct 2026

Parallel execution

RAMPART supports pytest-xdist parallel runs and produces a unified report across worker processes.microsoft.github.io · 8 Oct 2026

Security boundary

For xdist worker serialization, RAMPART documents JSON-safe payloads, rejection of unknown schema versions, a default 16 MiB per-result cap, and ANSI escape stripping.microsoft.github.io · 8 Oct 2026

Limitations

The xdist documentation says results recorded only during fixture teardown are excluded from report streaming and mixed RAMPART versions across controller and workers are unsupported.microsoft.github.io · 8 Oct 2026

Maker

The documentation identifies Microsoft Corporation as its copyright holder.microsoft.github.io · 8 Oct 2026

Test workflow

It orchestrates agent interactions, evaluates outcomes, and reports test results.microsoft.github.io · 9 Oct 2026

Attack testing

Its XPIA tests check whether malicious content planted in data sources can manipulate an agent when retrieved.microsoft.github.io · 9 Oct 2026

Behavior probes

Behavioral probes check for expected responses, tool use, or side effects and are described as useful for regression testing.microsoft.github.io · 9 Oct 2026

Evaluation output

Results include a safety status of SAFE, UNSAFE, UNDETERMINED, or ERROR, along with trace and evaluation details.microsoft.github.io · 9 Oct 2026

pytest integration

RAMPART registers as a pytest plugin automatically and provides harm and trial markers.microsoft.github.io · 9 Oct 2026

Reporting and CI

Built-in JSON report sinks and pytest-xdist support let teams collect reports for CI dashboards across parallel workers.microsoft.github.io · 9 Oct 2026

PyRIT integration

RAMPART uses PyRIT for LLM interaction, prompt normalization, and conversation memory, and installs PyRIT as a dependency.microsoft.github.io · 9 Oct 2026

Install requirements

The installation guide requires Python 3.11 or later and supports installation with uv or pip.microsoft.github.io · 9 Oct 2026

Optional integration

The optional onedrive extra adds dependencies for using the built-in OneDriveSurface to place XPIA payloads in OneDrive.microsoft.github.io · 9 Oct 2026

License

The GitHub repository identifies RAMPART as MIT-licensed.github.com · 9 Oct 2026

Intended users

The project describes its framework as a developer-friendly way to write and run safety and security tests for AI agents.github.com · 9 Oct 2026

Where it ranks on Inferse

• Best AI Security Testing Tools in 2026#14 of 29

Sources

• microsoft.github.io/RAMPART/· checked 8 Oct 2026
• microsoft.github.io/RAMPART/concepts/overview/· checked 8 Oct 2026
• microsoft.github.io/RAMPART/concepts/pyrit/· checked 8 Oct 2026
• microsoft.github.io/RAMPART/getting-started/installation/· checked 8 Oct 2026
• microsoft.github.io/RAMPART/usage/xdist/· checked 8 Oct 2026
• microsoft.github.io/RAMPART/attacks/xpia/· checked 9 Oct 2026
• microsoft.github.io/RAMPART/probes/behavioral/· checked 9 Oct 2026
• microsoft.github.io/RAMPART/usage/results-and-reporting/· checked 9 Oct 2026
• microsoft.github.io/RAMPART/usage/ci-integration/· checked 9 Oct 2026
• github.com/microsoft/RAMPART· checked 9 Oct 2026