Jonathan Mohrbacher johnnymo87

Tips for Using Claude Code Effectively

Getting Started & Configuration

Use the Best Model

Use Claude Opus 4.5 - it's the most capable model. If your company is paying for it, take advantage of it.

Run with --dangerously-skip-permissions

OpenAI's Financial Situation

Your brother mentioned hearing OpenAI isn't turning a profit. Here's what Zvi has written:

OpenAI's Losses and Cash Burn

From AI #133: America Could Use More Energy (Sep 11, 2025):

"OpenAI is now projecting that it will burn $115 billion (!) on cash between now and 2029, about $80 billion higher than previously expected."

Zvi's Take on Gemini 3 vs ChatGPT 5.1 (Nov-Dec 2025)

A summary of Zvi Mowshowitz's AI analysis for someone asking about the Tom's Guide "Gemini 3 vs ChatGPT 5.1" comparison article.

On Gemini 3 Pro

Zvi's headline: "Gemini 3 Pro Is a Vast Intelligence With No Spine"

Setup steps for Google Workspace MCP Server

Step 1

Create OAuth 2.0 credentials (web application) in Google Cloud Console
Navigate to APIs & Services > Credentials, then click Create Credentials > OAuth client ID.
Select the appropriate application type (Web application) and configure any necessary settings like redirect URIs or JavaScript origins.

Step 2

To the OAuth 2.0 credentials form, add redirect URI: http://localhost:8000/oauth2callback
Click create

Below is a 2025-focused playbook you can share with newcomers who want to “get their hands into a big code-base, learn fast, and land that first dev role.” In short: open-source is still the best proving ground, but the on-ramp is very different from when you broke in (2014). Cloud dev environments, Discord communities, AI pair-programmers, and funded mentorships mean you can contribute product-grade code in hours—not weeks—and get real humans vouching for you along the way.

1 Find projects that welcome newcomers

1.1 Use “Good First Issue” aggregators

goodfirstissues.com surfaces beginner-friendly tickets across thousands of repos; filters by language and topic ([goodfirstissues.com][1])
GitHub’s built-in good-first-issue topic now lists ~2 000 actively maintained repos, many from Microsoft, Mozilla, and Kubernetes ([github.com][2])

Development Environment Setup

NB: Assumes OS X

FES consumes events from the fes-async-in and the sts-delivery-tracking-events kafka event streams. The latter has never posed a problem, so this document is geared to talk about the former. Regarding fes-async-in, our order management systems send and receive kafka messages through it to communicate with our fulfillment management system. Prior to the solution, we in FMS would occasionally receive a kafka message that we couldn’t consume due to a bug, bad data, etc. When this happened, our kafka consumer tried this message again and again, literally forever. This “just try again” policy is great for transient errors, such as database locking. However, it's a problem if this error isn’t transient, as all other messages behind it are now blocked and will never be processed. This is what we call a "stuck message".

Inspired by this Uber blog post, we wrote [special "punt" code](https://github.com/blueapron/fulfillment-engine/pull

	#!/usr/bin/env bash

	: <<'END'
	Script Name: madge_concatenator.sh

	Purpose:
	This script uses 'madge' to find all dependencies for a given TypeScript/JavaScript
	entry file, and then concatenates all found files into a single output file.
	The output file will contain the absolute file paths and contents of each code file,
	separated by a delimiter (```). This is useful for providing a large codebase

	#!/usr/bin/env bash

	# Kafka Partition End Offset Checker
	# ==================================
	#
	# This script queries a Kafka cluster to retrieve the end offset information for all partitions
	# of a specified topic. It provides the log end offset for each partition, which represents
	# the offset where the next message would be appended.
	#
	# Usage:

	#!/usr/bin/env bash

	: <<'END'
	Script Name: code_concatenator.bash

	Purpose:
	This script is designed to concatenate all code files within specified directories
	and their subdirectories, or specified individual files, into a single output file.
	The output file will contain the file paths and contents of each code file,
	separated by a delimiter (```). This script is particularly useful for preparing