JevMade Sign in
← Back to guides

JevMade field notes / Integration walkthrough

Let an AI assistant choose safe actions from a strict list

This guide explains how to force an AI assistant to pick from a strict list of safe actions. It shows how to check the risk of a chosen task before letting the software actually perform it.

Original by ComposioIntegrations

Listen to this guide

JevMade’s plain-English explanation

0:00 /

Our summary

This guide shows how to connect an AI assistant to outside services using a tool collection called Composio. It uses Jev, an AI tool that chooses from options rather than writing an answer. This helps prevent the software from inventing unsafe actions.

When a person asks for help, the software looks at the available tasks. It must pick one exact match from the list or decide to do nothing. If it picks a task, a safety check runs before the software actually carries out the action.

This setup is useful for developers who want to block harmful changes. The example only allows actions that read information. You need private access codes for Composio and TypeSafe to try this. It does not guarantee every available action is completely safe.

Key takeaways

  1. Force the AI assistant to pick from a strict list instead of inventing its own actions.
  2. Always allow the software to safely skip a task if it cannot find the right match.
  3. Check the safety level of every chosen action before letting the software actually perform it.

This demonstration does not prove that every available tool is completely safe or labeled correctly. Anyone building this must create and enforce their own rules for what the software is allowed to do.

GitHub README and provider demo

Read the original guide Opens the author’s site in a new tab.

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.