Finance Coach Skill
Claude Code skill that turns personal finance data into a forward-looking dashboard — projections, scenario modeling, FI tracking, and investment thesis workshops.
Blanchett (2014) + BLS / EBRI Empirical Curve: Real household spending declines in mid-retirement before late-life healthcare rises — meaning our flat 3% rule was 56% over-provisioned.
The scariest bug in my personal finance tool was a 10-year net-worth chart that looked completely reasonable while keeping earned income running forever past retirement. Because the curve looked plausible, I didn’t question it until I audited the year-by-year table — and fixed it by adding a blunt survival line (“money lasts until age X”) that forces every projection to make a checkable claim.
I built Finance Coach as a Claude Code skill connected to live account data because standard finance apps stop at showing what you spent last month, not what happens over twenty years if you take a sabbatical, buy a home, or stress-test an investment thesis. It routes across five modes — current snapshot, multi-decade scenario charts, an investment workshop where a Challenger prompt names specific conditions that would prove a thesis wrong, advisor prep, and tax planning.
Building it showed how easily AI makes shaky inputs look rigorous. Before connecting real account data, I spent two sessions reviewing elaborate sensitivity tables built on a mental-math spending estimate that missed our actual number. Later, bumping annual spending growth from 2% to 3% flat to “be conservative” overshot how household spending actually changes with age by 56%, once I had Claude check federal Consumer Expenditure Survey data and retirement researcher David Blanchett’s age-curve studies.