GP's AI Research

Mixed Keyboard Shortcuts

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Thu, 20 Aug 2026 00:00:00 +0000

TL;DR: A cheat sheet of keyboard shortcuts across tools.

Vim Shortcuts

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Thu, 20 Aug 2026 00:00:00 +0000

TL;DR: A cheat sheet of Vim commands and concepts.

FastAPI in 60 Minutes

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Mon, 10 Aug 2026 00:00:00 +0000

TL;DR: Learn how to build a REST API with FastAPI and uvicorn in 60 minutes with hands-on examples covering request validation, dependency injection, and a complete Book Catalog API served over real HTTP

GitHub CLI Shortcuts

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Wed, 15 Jul 2026 00:00:00 +0000

A guide to useful GitHub CLI commands

TIL: Auto-Reload Files in Vim When Working With AI Agents

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Wed, 24 Jun 2026 00:00:00 +0000

TL;DR A 3-line vimscript snippet that polls checktime every second keeps vim buffers in sync when AI agents modify files on disk.

TIL: How to Not Let Coding Agents Commit Code Automatically

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Tue, 23 Jun 2026 00:00:00 +0000

TL;DR Coding agents shouldn't auto-commit. There are four ways to enforce that, from a static permission deny to a dynamic per-session authorization flag. The last one also needs a second, independent rule or the agent can create its own authorization.

My AI Policy

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Mon, 22 Jun 2026 00:00:00 +0000

TL;DR: I use LLMs as human team members and despise AI slop

Markdown Viewers on macOS: Terminal and Browser Options

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Mon, 22 Jun 2026 00:00:00 +0000

TL;DR: Render markdown from terminal using glow (for terminal viewing), mdcat (for rich formatting), grip (for GitHub-accurate previews with images), or pandoc (for document conversion and flexibility).

Rendering Markdown Files From Your Terminal with open_md.py

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Sun, 21 Jun 2026 00:00:00 +0000

TL;DR: open_md.py is a flexible CLI tool that renders markdown files to HTML using pandoc or grip, and can open them directly on GitHub or in your browser, with support for embedded diagrams and both local and dockerized execution

How to Use OpenRouter with Simonw's LLM CLI

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 12 Jun 2026 00:00:00 +0000

Guide on using OpenRouter with Simon Willison's LLM CLI

How to Use Claude Code with OpenRouter

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 05 Jun 2026 00:00:00 +0000

TL;DR Learn how to configure Claude Code to route LLM calls through OpenRouter, enabling access to models from multiple providers through a single API.

What Is LLM Reasoning Effort and How Does It Affect Cost?

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 29 May 2026 00:00:00 +0000

TL;DR: LLM reasoning effort settings (low, medium, high) affect how much compute a model allocates to internal reasoning, increasing total token consumption and cost, even when per-token pricing stays the same.

How to Count LLM Usage Costs

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 22 May 2026 00:00:00 +0000

Tools for tracking LLM API usage costs

How to Compare and Choose LLM Models

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 15 May 2026 00:00:00 +0000

TL;DR Comparing LLMs across real benchmarks and use an efficiency metric that combines quality, cost, and speed to pick the right model for each task

LLM CLI in 30 mins

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 08 May 2026 00:00:00 +0000

Transform text files with LLMs from the command line, integrate AI into your editing workflow

llm CLI in 30 minutes

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 01 May 2026 00:00:00 +0000

TL;DR: Simon Willison's llm CLI turns any large language model into a composable Unix tool you can pipe, script, and template from the terminal.

mdm, a Unified Markdown Manager for Agent Skills

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 24 Apr 2026 00:00:00 +0000

TL;DR: mdm unifies management of research ideas, blog posts, and Claude Code skills across multiple repositories into one powerful command-line tool.

Python Code Coverage In 30 Mins

saggese@gmail.com, gsaggese@umd.edu (Giacinto Paolo Saggese) — Fri, 17 Apr 2026 00:00:00 +0000

TL;DR: Python code coverage measures how much of your source code is tested, helping you improve test quality.

CausalML in 60 Minutes

TBD (Mohammed Syed) — Fri, 10 Apr 2026 00:00:00 +0000

TL;DR: Learn how to estimate heterogeneous treatment effects using CausalML in 60 minutes with hands-on examples including individual-level causal inference on CDC diabetes health data.

BambooAI in 60 Minutes

TBD (Aver81) — Fri, 03 Apr 2026 00:00:00 +0000

TL;DR: Learn how to build intelligent data analysis systems with BambooAI in 60 minutes with hands-on examples covering multi-agent workflows, dataframe analysis, and dynamic code generation.