RAMPART

Spec sheet

Purpose

RAMPART is a pytest-native safety testing framework for agentic AI applications.Read 9 Oct 2026 · microsoft.github.io

Intended users

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

License

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

Maker

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

Attack and probe tests

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

Attack testing

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

Behavior probes

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

Evaluation output

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

Included tests

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

Install requirements

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

Integration model

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

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.Read 8 Oct 2026 · microsoft.github.io

OneDrive

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

Optional integration

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

Parallel execution

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

PyRIT

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

PyRIT integration

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

pytest integration

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

pytest plugin

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

Python requirement

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

Reporting

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

Reporting and CI

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

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.Read 8 Oct 2026 · microsoft.github.io

Supported installation systems

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

Test workflow

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

Where Specifiction places it

• AI Security Testing Tools#21 of 297.3