How-To Guides
From measuring your first cascade to improving your yield — step-by-step guides for every stage of AI operator efficiency.
How to Measure AI Coding Efficiency
Measure your AI coding efficiency using the four token pillars and the Υ Yield metric with ccusage and sigrank. A privacy-preserving, on-device workflow.
How to Improve Your AI Coding Yield
Seven strategies to increase your Υ Yield: better context windows, prompt caching, structured inputs, fewer re-rolls, cache reuse patterns, and more.
How to Reduce Token Waste
Identify and fix the four common sources of token waste: repeated context, poor prompt caching, verbose prompts, and unnecessary re-rolls.
How to Read Your Token Cascade
Examine the four token pillars to diagnose your AI coding workflow. What high cache-read, high input/low output, and balanced cascades reveal.
How to Track Your Token Cascade
Install sigrank to automatically track the four token pillars from your local AI coding logs. Privacy-preserving, on-device, signed submissions.
How to Benchmark Your AI Coding Workflow
Establish a baseline yield, track the four token pillars across time windows, and compare yourself against the SigRank leaderboard.
How to Compare AI Operators
Use the SigRank compare tool to benchmark operators head-to-head on yield, cache hit rate, leverage, and class tier.
Cache Write Convergence — Troubleshooting
When ChatGPT/Codex reports cache_write as zero, use reference operating ratios to split the combined input. Validate against real operator data.