Welcome to Prompty

What is Prompty?

Prompty is an asset class and format for LLM prompts designed to enhance observability, understandability, and portability for developers. A .prompty file combines structured YAML frontmatter with a markdown prompt body — making prompts versionable, testable, and executable across languages.

---
name: greeting
model:
  id: gpt-4o-mini
  provider: openai
  connection:
    kind: key
    apiKey: ${env:OPENAI_API_KEY}
inputs:
  - name: userName
    kind: string
    default: World
template:
  format:
    kind: jinja2
  parser:
    kind: prompty
---
system:
You are a friendly assistant.

user:
Say hello to {{userName}}.

Python
TypeScript

import prompty

result = prompty.invoke("greeting.prompty", inputs={"userName": "Jane"})
print(result)  # "Hello Jane! How can I help you today?"

import { invoke } from "@prompty/core";
import "@prompty/openai";

const result = await invoke("greeting.prompty", { userName: "Jane" });

Key Features

One file, complete prompt — model config, inputs, tools, and instructions in a single .prompty file
Multi-language runtimes — Python and TypeScript today, C# coming soon
Pipeline architecture — render → parse → execute → process, each stage swappable
Agent mode — built-in tool-calling loop with error recovery
Structured output — define output schemas, get typed JSON back
Streaming — chunk-by-chunk delivery with full tracing support
Pluggable tracing — console, JSON file, or OpenTelemetry backends
VS Code extension — syntax highlighting, autocomplete, run & debug, trace viewer

Quick Links


🚀 Getting Started	Install, write your first prompt, run it
📖 Core Concepts	File format, pipeline, connections, tools, tracing
📋 Schema Reference	All frontmatter properties (auto-generated)
🔧 How-To Guides	Practical recipes for common tasks
🐍 Python	Python runtime guide
📘 TypeScript	TypeScript runtime guide

Contributing

Prompty is open source. We welcome contributions to the runtimes, tooling, and documentation. See the Contributing guide to get started.