Archive
Search
- I Built an Open Source MCP To Analyze Your Substack Data With Hermes And Codex A step-by-step setup and five prompts for asking useful questions about your publication without pasting private exports into a chat Dan Cucolea
- What Hermes Thinks I Do When I'm Not Talking To It My job title says one thing. The files I leave behind tell a much messier story. Dan Cucolea
- The One File That Made Hermes Finally Learn From Its Mistakes Persistent memory keeps useful context. A small ledger beside each workflow keeps the verified lesson where the next run can find it. Dan Cucolea
July 2026
- How Pangram Decides What Looks Like AI From its training data to the percentage on your screen, this is how Substack’s new AI detector works. Dan Cucolea
- How I Built A Substack API With Hermes And Codex I wanted Hermes to schedule my Substack Notes. A missing scheduling feature turned into an open-source tool built across Hermes, Codex, and a set of independent checks. Dan Cucolea
- OpenAI Is Coming for Hermes One Codex Update at a Time OpenAI keeps folding more of my agent stack into a $20 subscription. I still run Hermes as I want complete control over my workflows and the freedom to choose any model. Dan Cucolea
- The Twenty Minute Audit That Found Where Hermes Was Getting Lost I kept upgrading my stack and Hermes kept getting confused. Twenty minutes looking at the right layer told me why. Dan Cucolea
June 2026
- Tear Down Your AI Workflow and Rebuild It Like This Most AI workflows produce 80% noise because people tend to connect tons of systems before validating them against what their work needs. Here's the one test that saves you days of configuration. Dan Cucolea
- Why My Best Hermes Workflow Consists Of Plain MD Files I reorganized my vault so my agent could find things in thirty seconds. Boring fix, and it worked. Dan Cucolea
- I Built My Hermes Stack to Survive Vendor Lock-in and Google Just Proved Me Right Google is shutting down an open source tool in two days. Anthropic disabled Fable 5 because of the US government. My setup doesn't care, and here's why yours shouldn't either. Dan Cucolea
- How to Build AI Workflows When You're Tired of Optimizing Prompts It's time to move from asking questions to designing pipelines, and why your AI stops being useful when you need it most. Dan Cucolea
- Why AI Benchmarks Fail Real Hermes Agent Workflows AI benchmarks measure clean tasks. Agent workflows test messy realities. Here is how I route Opus, GPT, Qwen, and GLM across a real Hermes workday. Dan Cucolea
May 2026
- My Hermes AI Agent Maintenance Routine For Maximum Reliability Copy-paste prompts for checking Hermes memory, gateways, scheduled jobs, model IDs, and backups before silent drift breaks your agent workflows. Dan Cucolea
- I Tested 6 AI Plans to Find What $5, $10 and $20 Get You I ran six plans through Hermes to find out which ones finish the job and which ones just look cheap on the invoice. Dan Cucolea
- When to Use MCPs, CLIs, or Your Own Tool A practical decision framework for choosing between MCPs, CLIs, and custom wrappers to give Hermes, or any other AI agent, the right interface instead of connecting every tool. Dan Cucolea
- How to Use Claude Code For Free With OpenCode Models I built a Cloudflare Worker that lets Claude Code talk to OpenCode Go and Zen models, including free models like MiniMax M2.5 and Nemotron 3 Super. Dan Cucolea
April 2026
- How to Add Approval Gates to Your Hermes Agent A practical guide to adding human checkpoints to Hermes workflows, with examples for outbound content, system changes, and paid actions. Dan Cucolea
- How My Hermes Agent Plans My Morning Before I Have My Coffee Set up Hermes in 20 minutes and wake up to sorted tasks and curated articles instead of manual scrolling. Dan Cucolea
- The Agentic Engineering Shift Karpathy named two extremes. Here is the spectrum that covers everything in between, and the habits that shift you in the right direction. Dan Cucolea
- The $30 Hermes Stack That Makes Claude Max Look Like a Ripoff Claude charges $200 per month for Cowork and Code, but it limits you mid session. Here’s the Hermes stack that beats both with unified system memory, file sync, blazing fast speed, and unlimited usage Dan Cucolea
March 2026
- Hermes Is the AI Agent OpenClaw Promised to Be Hermes is the AI agent I’d recommend to anyone starting today. Here’s how I migrated from OpenClaw. Dan Cucolea
- Ditch Your Subscriptions and Run Open Source AI on Your Device Qwen 3.5 and MiniMax M2.5 cover every hardware tier from 2 GB to 128 GB. Here's how to match your machine. Dan Cucolea
- OpenClaw vs Claude Cowork vs Perplexity Computer - Which AI Agent Actually Fits Your Life Three tools, three philosophies, one guide to help you choose. Dan Cucolea
- Forgetting to Forget - How Infinite Memory Reinvents OpenClaw OpenClaw finally remembers everything. Here's what changed, how I set it up, and what it means for anyone building with AI agents. Dan Cucolea
- OpenClaw or Claude Cowork? Here's How to Plug Both Into the Same Brain The self-hosted architecture that lets OpenClaw and Claude Cowork read and write from the same files in real time. Dan Cucolea
February 2026
- The 'First Dollar' Stack The First Dollar stack: Why React + Supabase + Cloudflare is the only V1 combo that doesn't waste your weekends. Dan Cucolea
- How to Install and Configure PicoClaw on the Cheapest VPS The lightweight OpenClaw alternative claims it can run on a $10 Raspberry Pi. I tested it on the cheapest server I could find. Dan Cucolea
- The Armchair Expert Problem in Product Validation Don't let Claude be your Product Manager. Here’s how to save weeks of wasted building by swapping AI assumptions for a 30-minute "Validation Sprint.” Dan Cucolea
January 2026
- How to Get from Blank Screen to First Win in Under 60 Seconds Users can't tell the difference between "empty because I just started" and "empty because broken". The Activation Path fixes that. Dan Cucolea
- What Gordon Ramsay Taught Me About Vibe Coding If the user can survive without it, delete it. How to ruthlessly cut 50% of your feature ideas before writing a single prompt. Dan Cucolea
- The Build vs Buy Scorecard AI always tried to build things from scratch. This 60-second scorecard highlights the two paths it rarely mentions and helps you decide what to do: build, buy, or open source. Dan Cucolea
- 26 Mental Models to Build Better Products in 2026 Your bookmark-worthy reference guide for product thinking all year long. Dan Cucolea
December 2025
- Accepting AI's First Answer Is Bad For Your Code When AI suggests an approach, your immediate thought should be “What else exists?” instead of “Let me implement this.” Dan Cucolea
- Why AI Gets Stupid Deep in the Thread Why fresh sessions outperform long threads, according to new research from Chroma and Anthropic. Dan Cucolea
- How to Prompt AI for Consistent JSON Responses AI JSON fails silently in prod. Fix it forever with these 5 checkpoints. 🚀 Dan Cucolea
November 2025
- How To Architect A Feature In 5 Minutes Before Talking To AI AI code works on the surface but can be expensive underneath. Here is the architecture framework that fixes it. Dan Cucolea
- Your Landing Page Is Probably Over-Engineered Here's how to stop losing visitors to slow load times by serving instant HTML with Astro instead of React. Dan Cucolea
October 2025
- How to Search Beyond the Docs When Your Code Won't Work I wasted 45 minutes Googling a TanStack bug. Then Exa found the GitHub discussion with my exact fix in 5 minutes. Here's my setup. Dan Cucolea
- Kilo Code Changed How I Write Code and I'm Never Going Back No subscription. No credits that expire. No separate backend billing. Just pay for the AI you actually use, at the rates the providers actually charge. Dan Cucolea
- I Learned React in 9 Months and Built a Real SaaS - Yahini This journey starts with a quote I received from a developer to build the first version of Yahini. It was $15,000. Dan Cucolea