Tool Call
Pemanggilan Tool
The model's request naming a tool and its arguments. Structured output, not a side effect.
A tool call is the model naming a tool and supplying its arguments. The model produces it as structured output; the harness decides whether to run it. The call itself changes nothing.
This distinction is worth keeping. A model asking to delete a file is not a deleted file. The harness is where permissions are checked, arguments are validated, and the actual side effect happens. Every safety property in an agentic system lives in that gap between the request and the execution.
It also explains why a model can loop on a failing tool. If the call returns an error the model does not understand, the model often tries the same call again, because from its point of view it never got a result.
Related terms
Tool
A function the model is allowed to call, like reading a file or running a command.
Tool Result
What the harness sends back after running a tool: file contents, command output, an error.
Harness
Everything around the model that turns it into an agent: tools, system prompt, permissions, hooks.
Agent Loop
The repeated cycle of decide, act, observe. Where autonomy actually comes from.
Guardrails
Checks that stop or constrain an action before it happens.
Want this working inside your team?
It starts with one conversation about your workflow.