Heads up! To view this whole video, sign in with your Courses account or enroll in your free 7-day trial. Sign In Enroll

Add Guardrails and Safety to Your AI Agent - AI Agents Day 25

Well done!

You have completed Add Guardrails and Safety to Your AI Agent - AI Agents Day 25!

Sign up for Treehouse Back to Library

Secure Your Agent Against Prompt Injection

16:59 with Daniel Montgomery

Run a security audit against your agent, then work through the findings. Add authentication, sender allow lists, and approval controls, then install a guard layer that screens web content and recalled memory before either reaches the model.

Click here to download Eve Agent Security hardening doc

Resources

Related Discussions

Have questions about this video? Start a discussion with the community and Treehouse staff.

Sign up

    Related Discussions

    Have questions about this video? Start a discussion with the community and Treehouse staff.

    Sign up

    Next, we want to make sure our agent is as secure as possible. 0:00
    We want to make sure since this is a local agent, that only we can run it, and it 0:04
    ignores all the outside requests if there are any. 0:08
    We also want to stop any prompt injection attempts as well. 0:11
    Prompt injection can occur when we send our agent to the web to scrape and find information. 0:14
    Some websites can have malicious prompts or code hidden in them that they'll use in 0:20
    an attempt to override our agent. 0:25
    And since we have persistent memory that can also add memories to the database on 0:27

    You need to sign up for Treehouse in order to download course files.

    Sign up

      You need to sign up for Treehouse in order to set up Workspace

      Sign up