> ## Documentation Index
> Fetch the complete documentation index at: https://docs.klatt.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Tool lifecycle

> Give an agent a bounded action, configure its inputs and authentication, then verify its success and failure behavior in a test call.

Give an agent a bounded action, configure its inputs and authentication, then verify its success and failure behavior in a test call.

## Prerequisites

* A workflow with an Agent node.
* A specific action that should be available to the agent.

## Add and verify a tool

1. Open **Tools** and choose **Create tool**.
2. Select the tool category and give it a clear name and description.
3. Configure the destination, parameters, credentials, or call behavior for that category.
4. Attach the tool to the Agent node that should invoke it.
5. Test a valid invocation, a rejected input, and a timeout or unavailable destination.

## Constraints and failure states

Only configured tools attached to the current node are available during that exchange. A tool can return an error without ending the call; the prompt should tell the agent how to recover or ask for clarification.

## Protect your data

Use least-privilege credentials and validate tool inputs at the destination. Never place a provider secret in a tool description or a spoken prompt.

## Related

* [HTTP tool](/voice-agent/tools/http-api)
* [End Call tool](/voice-agent/tools/end-call)
* [MCP connections](/integrations/mcp)


## Related topics

- [HTTP tool configuration](/voice-agent/tools/http-api.md)
- [End Call tool](/voice-agent/tools/end-call.md)
- [MCP connections](/integrations/mcp.md)
- [Call transfer tool](/voice-agent/tools/call-transfer.md)
- [Pre-recorded audio](/voice-agent/pre-recorded-audio.md)