Tools, Permissions, and Knowledge Access

This article explains how tools, permissions, and knowledge access affect how a custom agent behaves in Moterra. Choosing the right configuration helps keep custom agents reliable, safe, and focused on their intended task.

Written By Kristė Vagnerytė

Last updated 22 days ago

What Tools Control

Tools define what a custom agent is allowed to do.

Only enable the tools that are necessary for the agent’s purpose. Fewer tools usually lead to more predictable results.

Custom agents can use the following tools:

  • Artifacts Toolkit – Create and edit structured documents

  • Charts – Create charts and visualizations from data

  • Database Analysis Toolkit – Query and analyze connected databases

  • Deep Research Toolkit – Perform deeper research using external sources

  • Image Generation Toolkit – Generate images from descriptions or prompts

  • Knowledge Base Toolkit – Retrieve information from the internal knowledge base

  • Translation Toolkit – Translate documents or text between languages

  • Weather Toolkit – Get current weather and forecasts for a location

  • Web Search / Scrape Toolkit – Search and retrieve up-to-date information from the web


How Permissions Affect Agent Behaviour

Permissions control who can see and use a custom agent.

You can use permissions to:

  • Limit access to specific users or groups

  • Prevent sensitive agents from being used broadly

  • Ensure agents are used only in the right context

Well-defined permissions help reduce misuse and confusion.


How Custom Agents Use Knowledge

All custom agents currently use the same knowledge base.

How an agent uses that knowledge depends on:

  • Its system prompt

  • The tools that is allowed to use

  • Its defined scope

Clear instructions help the agent know when and how to rely on internal knowledge.


Best Practices for Configuration

Follow these best practices when configuring tools and permissions:

  • Enable only the tools the agent needs

  • Keep permissions as limited as possible

  • Align tools with the agent’s scope

  • Review configurations when agent behaviour changes


What to Read Next