What problem does it solve? Manual prompt engineering is brittle, hard to maintain, and difficult to improve systematically. This Skill replaces hand-tuned prompts with declarative, data-driven LM programming using Stanford NLP's DSPy framework, so prompts and pipelines optimize automatically from training examples. ## Core Features & Use Cases - Declarative Signatures & Modules: Define tasks as input-output signatures and compose them with Predict, ChainOfThought, ReAct, and ProgramOfThought modules. - Automatic Prompt Optimization: Use teleprompters like BootstrapFewShot, MIPRO, and BootstrapFinetune to improve prompts and generate fine-tuning data from metrics. - Composable Pipelines: Build multi-stage RAG systems, agents, classifiers, and extraction pipelines with typed outputs via Pydantic. - Use Case: Build a multi-hop RAG question-answering system, define an exact-match metric, then compile it with BootstrapFewShot on 50 examples to measurably improve answer accuracy without rewriting any prompts. ## Quick Start Use the dspy skill to build a chain-of-thought question answering module and optimize it with BootstrapFewShot on my training examples.
