Build AI pipelines with declarative signatures and automated prompt optimization.
npx skills add https://github.com/cryptopafi/nexusos-skills --skill dspy-cryptopafi
Please help me install this Agent Skill. Skill: dspy Source: https://github.com/cryptopafi/nexusos-skills/tree/main/mlops/research/dspy Command: npx skills add https://github.com/cryptopafi/nexusos-skills --skill dspy-cryptopafi
Building robust, modular AI systems often requires brittle manual prompt engineering, ad-hoc chains, and time-consuming tuning; dspy provides a declarative layer and automated optimization to make LM-based pipelines predictable, maintainable, and improvable with data.
Use dspy to define a signature and assemble a RAG pipeline that retrieves documents, composes context, and returns concise answers.
High-intent search queries and answers about installing and using this skill.
Programmatic prompt optimization uses automated optimizers like BootstrapFewShot and MIPRO to iteratively improve prompts and demonstrations from data. This replaces manual tuning by compiling declarative signatures into optimized instructions and few-shot examples.
Declarative LM programming defines AI tasks as signatures and composes reusable modules like Predict and ChainOfThought. This approach abstracts brittle prompt engineering, allowing developers to build maintainable and predictable multi-stage language model pipelines.
Retrieval-augmented generation agents are supported through modular primitives like ReAct and ProgramOfThought. You can assemble pipelines that retrieve documents, rerank passages, and generate answers with integrated evaluation utilities for production reliability.
Local models, Anthropic, and OpenAI integrations are fully supported through native LM provider configurations. You can switch between providers by configuring the language model client, enabling flexible deployment for research and product teams.
Automated few-shot optimization should be used when building classifiers or pipelines that require predictable, data-driven improvements. Optimizers like BootstrapFinetune generate reliable demonstrations from labeled data, eliminating ad-hoc prompt chains and manual tuning.
The best way to structure multi-stage AI workflows is using declarative modules to compose tasks. By defining signatures and applying automated optimizers, you replace brittle ad-hoc chains with maintainable, improvable pipelines that include batching, caching, and monitoring.
gztcodeProgram language models declaratively with automated prompt optimization in DSPy.
stephanjArchitect RAG systems integrating vector databases, embedding models, and retrieval pipelines.
zerwizBuild declarative LLM programs and RAG pipelines with DSPy.
guccangOptimize language model programs with declarative task definitions and automated prompt improvement.
