The workshop's notebook
Guides
Practical references with a verification date on every page. Written next to the tools they feed, refreshed on a schedule — if a number is on this shelf, it was checked against the primary source, not remembered.
LLM & Structured Output
How Much Does a Multi-Turn LLM Conversation Cost?
Why a chat or agent costs far more than a single prompt — the full history is resent every turn so cost compounds — and how prompt caching cuts it 70-90%, with a worked example.
LLM & Structured Output
LLM API Pricing Comparison (July 2026)
All eleven current models in one table — GPT-5.6, Claude 5 family, Gemini 3 line — plus what three real workloads cost per month at list price.
LLM & Structured Output
Structured Outputs with JSON Schema
The five schema rules that hold across GPT, Claude and Gemini — strict mode, nullable-not-optional, enums — plus the repair-and-retry loop.
Regex & Patterns
Cron Expression Examples — 30 schedules
Every 5 minutes, weekdays at 9, first of the month — grouped by job, with the DOM-vs-DOW OR rule and the timezone pitfalls.
Metadata & SEO
OG Image Size & Social Meta Tags
1200×630 and the five tags that cover 95% of shares — truncation limits per platform and cache-debug moves when previews refuse to update.
Calculators
99.9% Uptime = How Much Downtime?
The full nines table per day/week/month/year, error budgets, and why composite SLAs multiply against you.
JSON & Config
JSON Patch vs Merge Patch
RFC 6902 operation lists vs RFC 7386 partial documents — the null-means-delete trap, array semantics and the decision table.
Calculators
SaaS Metrics Formulas + Benchmarks
Nine formulas on one worked example — margin-adjusted LTV, CAC payback, quick ratio — with the 3:1 / <12mo / >75% thresholds.