Freebies
Immediate Value Delivered
Free things you can use today: the tactics I lean on most, the open tooling behind how I build, and small tools that do one useful thing well. No signup wall in the way.
Tactics
Tactics
Short, do-it-now moves I lean on. Read one, use it the same day.
The Four-Part Prompt Skeleton
The context, task, constraints, and format structure I fall back on for almost any prompt. Copy it, fill in the blanks, and get usable output on the first try instead of the fifth.
Open it
The Parlor-Trick Test
A short set of questions I run any AI demo through before I believe it. Separates something that ships value from something that only looks good on stage.
Open it
Build vs Buy in Five Questions
The five questions I ask before subscribing to another SaaS tool. AI has moved this line toward building more than most people realize, and this is how I check.
Open it
Honest Metrics Starter Set
The handful of marketing signals I still trust after AI flooded the web, and the vanity numbers I stopped reporting. A calmer, more defensible way to measure.
Open it
Skills
Skills
The actual Claude Skills I run day to day, packaged so you can drop them straight into your own Claude Code setup.
Gauntlet Loop
The builder or critic loop I use for high bar builds: a lead agent splits a goal into pieces, each piece gets a separate builder and a fresh-context critic, and an architecture critic holds veto over the whole thing so per-piece polish cannot ship incoherent work.
Long-Haul Build
How I keep a long unattended coding run honest: a signed spec with numbered acceptance criteria, a verifier that runs real checks instead of reading the agent's own claims, and a separate critic that scores alignment to the spec, not just quality.
Coding SOP
My pre-commit checklist for AI-assisted coding: inspect the diff, run the real gates, stage on purpose, and stop and report instead of force fixing when something breaks.
Codebase Wiki Builder
A living wiki structure for a codebase, built and kept current as the code changes instead of going stale the week after launch.
Feature Planning
The right sized documentation package I write before building a feature, so implementation follows a spec instead of improvising against the ticket title.
Accessibility Audit
A structured WCAG 2.2 accessibility audit for any website or web app: what to check, in what order, and how to grade severity.
Accessibility Remediation
How I fix the accessibility violations an audit turns up, contrast, keyboard nav, ARIA, and focus management, without breaking the design in the process.
Mobile Responsiveness Auditor
Finds where a site breaks on real phones: the app type it is, current best practice for that type, then a pass over breakpoints and touch targets.
Production Hardening
Stress tests a build against the paths that do not show up in a demo: network failure, empty data, malformed input, slow connections, concurrent users.
Beta Readiness Audit
Grades an app against a real beta readiness bar and returns a prioritized gap list instead of a vague look good to me.
Agent Cost Profiler
Finds why an AI agent's token cost or latency is high and which lever, not every lever, actually fixes it.
Recursive Learning Loop
The loop I run after a real correction or a fail then recover: pull out what generalized, write it down as a reusable rule, and use it before re-deriving the same lesson twice.
Web Motion Design
Tasteful, accessible motion for a website, starting from the cheapest technique that would work and only escalating when the cheap one is not enough.
User Journey Tracer
Walks one user journey step by step from the user's side to find dead ends, missing transitions, and broken handoffs before a real user does.
Persona User Testing
Spins up a roster of persona based agents to act like real users on a deployed app and reports back where they got stuck.
Harness repos
Harness repos
The open tooling behind how I build. Public repos you can clone, read, and make your own.
Agent Starter Harness
A stripped-down version of the pattern behind Hub Helper Harry: retrieval with citations, memory, and model routing, ready to read and fork. The public repo is being prepared.
MCP Connector Examples
Working examples of wiring an agent into real tools with the Model Context Protocol, the connective tissue most tutorials skip. Public repo link pending.
Vibe Coding Project Template
The Next.js, Supabase, and Vercel skeleton I start real builds from, set up so a non-traditional engineer can ship a first version fast. Public repo link pending.
Tools
Tools
Small utilities that do one useful thing well, free and no signup wall.
Switching Cost Calculator
You enter what you pay now, what the alternative costs, and what moving would cost in fees and your team's hours, and it returns the monthly saving, the payback month, and where you net out at 12, 24, and 36 months. It runs entirely in your browser, it will tell you to stay when the numbers say stay, and it compares prices only, so it cannot tell you whether the new tool is any good.
Open it
What To Automate First
You list the repetitive tasks in your week, with how often they run and who does them, and it ranks which to automate first, what each one costs you a year in hours and dollars, and which to leave with a person. It runs entirely in your browser and it will tell you when a task is not an automation candidate at all. It cannot tell you how to automate anything or what building it would cost.
Open it
The Software Regret Check
You answer ten questions about the software deal you are about to sign, contract length, pricing model, data export, who set it up in motion, and it scores the regret risk and tells you what to negotiate before you commit. It runs entirely in your browser. It scores the terms of the deal, not the product, so it cannot tell you whether the software is any good.
Open it
Key Person Log
You capture the thing only you, or one person, knows how to do, the moment you notice it: what it is, what someone should do about it if you are not around, and how urgent it is to write up properly. It builds into a growing log you can copy as plain text and hand to a successor or paste into an onboarding doc. It runs entirely in your browser. It only holds what you type in this browser tab, so it cannot back your log up, sync it across devices, or write the real documentation for you.
Open it
The Free AI Assessment
Tell me about a role or workflow and I estimate what automating it could be worth, using public wage data and cited benchmarks, without ever needing your internal financials. Always labeled as an estimate.
Open it
Ask the Site Concierge
The chat on this site can run a short, no-hype discovery and give you a grounded read on where AI realistically fits for you. Open it any time from the corner of the screen.
Open it