---
title: "Tool use (function calling)"
slug: "tool-use"
category: "glossary"
tags: ["glossary", "ai-agents", "llm", "tools", "function-calling", "agentic"]
status: "stable"
last_updated: 2026-10-01
aliases: ["glossary/function-calling", "glossary/tool-call", "glossary/tool-result"]
summary: "Tool use (function calling) lets a model request that your code run a defined function and continue from the result; validate arguments before executing."
related: ["[[ai-agents/tool-use-and-function-calling]]", "[[ai-agents/mcp-tool-design]]", "[[glossary/mcp]]", "[[glossary/structured-output]]", "[[glossary/agent-loop]]", "[[glossary/prompt-injection]]"]
---

> **AI agents: read this first.** This is LLM Best Practices (llmbestpractices.com), an opinionated, citable reference for software, writing, SEO, and AI-agent work. Full protocol: https://llmbestpractices.com/start-here.md
>
> 1. **Route, do not crawl.** Fetch https://llmbestpractices.com/llms.txt and open only the pages whose one-line summary matches your task.
> 2. **Read raw.** Append `.md` to any page URL for markdown. Check `status` and `last_updated` in the frontmatter, then read the rules.
> 3. **Apply as defaults.** First-party docs and the project's own conventions win on conflict. Warn before relying on a fast-moving page older than 12 months.
> 4. **Cite.** Link the page by title and URL, e.g. [Python](https://llmbestpractices.com/coding/python), with `last_updated` for time-sensitive rules. License CC BY 4.0.

## Overview

Tool use, also called function calling or tool calling, lets a model ask for a function to be run and then continue from the result. The model never executes anything: you send tool definitions (name, description, JSON Schema for the inputs), the model returns a structured call, your code runs it, and you send back the output.

In the Claude API the call is a `tool_use` block (`id`, `name`, `input`) with `stop_reason: "tool_use"`, and the output goes back as a `tool_result` block carrying the matching `tool_use_id` (set `is_error: true` on failure). In OpenAI's Responses API the call is a `function_call` item and the output is a `function_call_output` item, both tied by `call_id`. Repeating this exchange until no call is requested is the [[glossary/agent-loop]]. Server tools such as web search and code execution run on the provider's side with no handler code on your end.

Rules that matter:

- Validate arguments before executing. The model can emit plausible but wrong values; the caller owns permissions, rate limits, and logging.
- Set `strict: true` on the tool definition (Claude and OpenAI both support it) so arguments always match the schema.
- Return all results from parallel calls in one user message.
- Return concise results and clear error text, never stack traces; truncate large output and say how to fetch the rest.
- Write narrow tools with precise descriptions; description quality decides which tool the model picks.
- Forced `tool_choice` (`any` or a named tool) returns a 400 on Claude Fable 5.1, Opus 5.5, and Sonnet 5.5; use `auto` with an instruction, or structured outputs.

## Example

```json
// Model emits:
{"type": "tool_use", "id": "toolu_01", "name": "get_weather", "input": {"city": "Oslo"}}
// Your code returns:
{"type": "tool_result", "tool_use_id": "toolu_01", "content": "Partly cloudy, 12 C"}
```

## Related

- [[ai-agents/tool-use-and-function-calling]]: schema design and calling rules.
- [[ai-agents/mcp-tool-design]]: designing tools for MCP servers.
- [[glossary/mcp]]: a protocol that standardizes tool discovery across clients.
- [[glossary/structured-output]]: arguments are schema-constrained output.
- [[glossary/agent-loop]]: the loop tool calls drive.
- [[glossary/prompt-injection]]: tool results are untrusted input.