dsl-spa 0.3.2

A tool for building Structured Agentic Pipelines

Domain Specific Language - Structured Pipeline Agent

DSL-SPA is an open-source Python library for connecting LLM Agents to ETL and other common tasks. With this tool, you can connect LLM Agents to complex Data Pipelines.

Key Features of DSL-SPA

• Building SQL Queries - construct sql queries from fields extracted by Agents

• Applying Data Transformations - selectively apply data transformations based on Agent instructions

• Generating Unstructured Text Summaries - generate summaries of query results and data transformations

• Generating Visualizations - generate vega-lite visualizations of query results and data transformations

• Executing Commands - generate command sequences or execute python fucntions

For some examples of what you can do with dsl-spa sdk check out the streamlit demo.

Getting Started

Get started with dsl-spa by installing the Python library via pip

pip install dsl-spa

A standard pipeline can be imported with

from dsl_spa.pipeline.pipeline import StandardPipeline

For more details on constructing a pipeline, reference Building a Pipeline.

For more details on the flow of Semantic Caching using DSL-SPA, refer to Semantic Cache Flow