function calling and tool use

[hugokoopmans]

I am struggeling to get a prompty to deal with more then one tool. How should i go about if i need to have more then one tool defined in my tools.json and handle these in the prompt so that in the conversation, if applicable the llm will sugest and use a specific tool?

I do not see how i can give more then one specific sample/example to a prompty?

Should i build a prompty for each tool / function call?

Maybe even more generic: how should i handle functioncalling in prompty in the first place?

In my usebase we talk about cars and CO2 emission, i have multiple tools that gather specific information about the car, make, model, engine, fueltype and then another tool to calculate emission class of the vehicle,,,

please advise,

thx hugo

[gaborvar]

@hugokoopmans I went down a very similar path. Happy to provide my experience. I added a handful of extra tool calls to the Azure-Samples/azure-search-openai-demo#2340 code base that you also use.

[hugokoopmans]

ok i wil dive in.... thx

[sethjuarez]

Hey @hugokoopmans! Prompty v2 has first-class support for multiple tools in a single prompty file. The new documentation covers this in detail:

• Tools (Core Concept) — how to define multiple tools (function, MCP, OpenAPI) in your frontmatter
• Agent Mode — how the automatic tool-call loop works (LLM picks the right tool, executes it, loops until done)
• Agent & Tool Calling How-To — step-by-step guide with Python and TypeScript examples

In v2, you define all your tools in the tools: section and the agent loop handles dispatching automatically — no need for separate prompty files per tool.

The site just deployed at https://prompty.ai/ — let us know if this helps with your use case!