Skip to main content
Tools are functions that a model can call during prompt execution. When the model decides it needs external information or wants to perform an action, it requests a tool call. PromptJuggler executes the tool, feeds the result back, and the model continues – potentially requesting more tools in the same run, across as many rounds as the conversation needs.

​
Attaching tools

Open a prompt’s configuration toolbar, click Tools, and add as many as you need. Each tool type has its own configuration. You can mix types freely – a prompt can have an HTTP call, a knowledge base search, and an MCP server tool all attached at once. Every tool has a name, a description (which tells the model what the tool does and when to use it), and a fail-fast toggle. When on, a failed tool call fails the entire prompt run. When off, the error is fed back to the model so it can try a different approach. Names must be unique within a prompt, and web_search is reserved for the built-in web search tool — it is the name search reports under in a run’s transcript, so a second tool sharing it would be indistinguishable to whoever is reading.

​
Tool types

Prompt & Workflow

Call another prompt or trigger a workflow as a tool. Turtles all the way down.

Knowledge Search

Search a knowledge base for relevant context.

HTTP Call

Make HTTP requests to external APIs.

Script

Run custom code in a secure sandbox.

Web Search

Search the web for current information.

MCP

Connect to Model Context Protocol servers.

Emit

Return schema-validated structured data beside the prose.

​
How tools work in workflows

When a prompt with tools runs inside a workflow, the tools execute as part of that prompt node’s run. Tool results are internal to the prompt – they feed back into the model’s reasoning but don’t flow through the workflow’s edges. The workflow only sees the prompt’s final output. If you want a tool’s result to be available to other nodes, use a dedicated node type instead: a KB Search node, a Script node, or wire the data through explicitly.