Skip to main content
Prompt engineering is difficult: minor variations in prompts have large impacts on the output of LLMs and prompts are model-dependent. In recent years prompt programming languages have emerged to bring discipline to prompt engineering. Many of them are embedded in an imperative language such as Python or TypeScript, making it difficult for users to directly interact with prompts and multi-turn LLM interactions. The Prompt Declaration Language (PDL) is a YAML-based declarative approach to prompt programming, where prompts are at the forefront. PDL facilitates model chaining and tool use, abstracting away the plumbing necessary for such compositions. It enables type checking of the input and output of models. PDL has been used with application patterns like RAG, CoT, ReAct, and an agent for solving SWE-bench. PDL is open-source. You can use PDL stand-alone or from a Python SDK. In a Jupyter notebook, a convenient extension lets you write PDL directly, without the need to write Python “boilerplate”. It even provides color coding of the YAML declarations and in the cell output, model-generated text is rendered in green font, and tool-generated text is rendered in purple font. You will use this notebook extension in this recipe.

Get started

Jupyter notebook introducing Prompt Declaration Language with Granite