Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.mindgard.ai/llms.txt

Use this file to discover all available pages before exploring further.

To familiarize yourself with Mindgard we recommend testing a demo AI model first. From the Overivew page click Let’s get started then click Run a test on a demo target on the Getting Started page. This will execute a suite of single turn attacks against a model hosted by us. The specific model currently in use will be clearly indicated next to the button. Run Demo Test Pn After a quick popup describing the steps to stand up our model and start the test, you’ll be redirected to the Projects page. Here you can review the results identified from any tests you have run. Your test should appear at the top of the page and include Demo Project in its name. Testing our model may take a few minutes to complete. Click into the demo project to see results as they are available and monitor testing progress. Tests List Pn At the top of the test overview will be a timeline of test results. If this is your first test, the timeline will only include a single result. Scroll to the bottom of the page to see each of the attack techniques deployed in the test and their results once available. Flagged events indicate risky responses from the test model. Test Overview Pn Click into one of the attack techniques to see detailed results. The next page shows you:
  • Summary: The number of attack attempts against the AI model and a count of those deemed successful.
  • Framework Mapping: The attack mapped to industry-standard frameworks such as OWASP Top 10 for LLMs and MITRE ATLAS.
  • Remediation: Recommendations to reduce the system’s susceptibility to this attack technique.
Attack Details
  • Event Details: Display the specific prompts used in each attack and the responses received. Due to the sensitive nature of some tests, event details will be obfuscated until explicitly clicked to view.
Event Details This is the standard workflow for running tests: execute a test focused on a goal and review results to craft your next steps. Developers will make changes to address risk, and test again to verify their work. Red teamers will develop new goals and test again to discover more risk. Continue to the next section to learn how to use Mindgard’s CLI to run tests against your own systems.