Skip to content

Giving an agent a new capability Guide

Ask the AI builder to give your agent a new tool or integration, and learn why a fresh agent can already use every built-in.

Last updated July 16, 2026

On this page

An agent's capabilities are everything it can actually do: the tools and integrations it can reach, the skills it follows, and the knowledge it reads. The simplest way to give an agent something new is to ask the AI builder for it in plain English.

How to add a capability with the builder

Every agent is set up and adjusted through Praxivara's AI builder, Stan, on the agent's Build tab. You describe what you want in ordinary language, and the builder wires up the right tools, integrations, and skills for you.

  1. Open the agent and go to the Build tab.
  2. Tell the builder what you want the agent to be able to do. Be specific about the outcome, for example: "Let it send WhatsApp messages," "Give it access to web search," or "Have it post updates to our Slack channel."
  3. The builder proposes the change and adds the matching tool, integration, or skill.
  4. Run a quick test to confirm the agent can now do what you asked.

Tip: Describe the goal, not the tool name. If you ask for the result you want ("email me a weekly summary"), the builder chooses and installs the right capability for you.

Why a new agent can already do a lot

An agent that has no explicit tools set can use all of Praxivara's built-in tools automatically. These are the everyday capabilities that don't require connecting an outside account, such as emailing you, saving results, searching the web, working with files, and generating images.

This means a brand-new agent is useful right away. You generally only need to add a capability when the agent should reach a service you connect yourself.

Built-in tools vs. integrations

It helps to know the two kinds of capabilities an agent can have.

  • Built-in tools are always available to the agent, with nothing to connect.
  • Integrations let the agent work inside your connected accounts, such as Gmail, Slack, or Shopify. An integration has to be connected before the agent can use it.

If you ask the builder for an integration the agent isn't connected to yet, it will guide you through connecting the account first.

See what an agent already has

You can review an agent's capabilities at any time from its detail page.

  1. Open the Tools tab to see every tool and integration the agent can use, grouped by integration, with the connected account shown for each.
  2. Open the Skills tab to see any installed skills. A skill is a proven, step-by-step playbook for a specific job, and installing one also grants the extra tools that skill needs. That's why an agent sometimes has tools you didn't add directly.

Reviewing these tabs is the quickest way to answer "why didn't the agent do that?" If a needed tool or integration isn't listed, ask the builder to add it.

Confirm and roll back changes

After the builder makes a change, use the agent's test run to check the new capability works as expected. Version history keeps a record of changes, so if a change isn't what you wanted, you can roll back to an earlier version.

When a capability needs a key or token

Some outside services require an API key or token to connect. When an agent needs one, the builder will ask you for it, or you can add it yourself on the agent's Secrets tab. Values you save there are encrypted and hidden after saving, and the agent references them securely by name rather than seeing the raw value.

Was this page helpful?