Shane Kercheval

llm-workflow. create workflows with LLMs

mcp-client-agent. CLI that uses DSPy to interact with MCP servers.

mcp-this. mcp-this lets you turn any command-line tool into an MCP tool and create structured prompt templates that any MCP Client (e.g. Claude Desktop) can use. er for any command

misc-projects.

shiny-explore-dataset. R

shiny-ab-testing. A shiny app for analyzing AB tests

chat-ai. Python

r-predictive-analysis-template. Template for finding the best potential predictive models for a given dataset, using a shotgun approach.

explore-data. A dash app that helps users explore data

flex-evals. A Python implementation of the Flexible Evaluation Protocol (FEP) - a vendor-neutral, schema-driven standard for evaluating any system that produces complex or variable outputs, from deterministic APIs to non-deterministic LLMs and agentic workflows.

flask-postgresql-template. Make it easy to set up a flask (w/ postgresql) project. Made to deploy on Heroku.

switchboard. Manual orchestrator for AI coding agents. Spawn Claude Code and Codex sessions, route messages between them, and define reusable fan-out / fan-in patterns. Built for developers who want explicit human-in-the-loop control over multi-agent workflows.

oo-learning. Python machine learning library based on Object Oriented design principles; the goal is to allow users to quickly explore data and search for top machine learning algorithm candidates for a given dataset

study-ai. Using Generative AI to aid in studying

reasoning-agent-api. An OpenAI-compatible API that adds reasoning capabilities and tool usage through remote MCP (Model Context Protocol) servers. Includes a simple web interface for conversations.

chat. Python

local-llm. Jupyter Notebook

tiddly. Python

flex-evals-viewer. Lightweight React app that views eval results generated from pytest & @evaluate