If this is useful, a ❤️ helps others find it. I run both in production. Here's the real comparison — not theoretical, from actual use building developer tools. Local LLM (Ollama) Gemini API (Free) Cost $0 forever $0 (free tier) Privacy 100% local Data sent to Google Setup Install Ollama + pull model Get API key (2 min) Quality Good (7B), Great (70B) Excellent Speed Fast if model lo
When you build a PowerShell project from multiple files, the natural structure is clear: enums first, then classes, then functions. Each group has its own place, and as long as dependencies only flow in one direction, that structure works perfectly. But sometimes a function depends on a class, and that class calls the function. There is no longer a clean boundary between the two groups — they need
The drift problem nobody told you about If you have used Claude Code, Cursor, Aider, or any other AI coding agent across more than two projects, you have felt this: You start project A. You copy the .agents/ folder (or CLAUDE.md, or .cursorrules) from your last project. You tweak two things. Done. You start project B six weeks later. You copy from project A. You tweak three things this time. Now
Cross-posted from the Stigmem blog. Today we're releasing stigmem v1.0: A stable, open-source specification and reference implementation for a federated knowledge fabric for AI agents. Stigmem = Stigmergy + Memory. Stigmergy (Greek stigma — mark; ergon — work) is the coordination mechanism you see in ant colonies and termite mounds: agents don't communicate directly with each other. Instead, they
More rules should mean better output. That's the intuition. I spent weeks building a comprehensive CLAUDE.md — 200 lines covering naming conventions, security rules, error handling, architectural patterns, import ordering, type safety requirements, and more. I was proud of it. I'd thought through every scenario. Then I scored the output. 79.0 / 100. My carefully crafted documentation was actively
If this is useful, a ❤️ helps others find it. I debug Rust and TypeScript code daily. I've used all three major AI APIs for this — Gemini, Claude, and GPT-4. Here's the honest comparison for code debugging specifically. Not benchmarks. Actual use. I ran the same 5 bugs through each model: A Rust borrow checker error with async context A React state update causing infinite re-render An Android logc
If this is useful, a ❤️ helps others find it. I've shipped 7 Mac apps in the past year. Every AI feature in them runs on free tools. Here's the exact stack — what I use, why, and where the limits are. What: Gemini 2.5 Flash Preview via REST API Cost: Free tier — 500 requests/day, no credit card Use for: Log diagnosis, document analysis, text classification, anything needing strong reasoning The fr
If this is useful, a ❤️ helps others find it. Everything I keep looking up when building with Gemini — in one place. Model Context Best for gemini-2.5-flash-preview 1M tokens General use, thinking, fast gemini-2.5-pro-preview 1M tokens Complex reasoning, best quality gemini-1.5-flash 1M tokens Stable, production-ready gemini-1.5-pro 2M tokens Longest context gemini-2.0-flash-lite 1M