# Retire Runway

> Will the money last? A household's retirement plan projected free in the browser - a year-by-year cash flow in
> today's dollars, a 1,000-path Monte Carlo probability of success against an 85% target, the stress test and Social
> Security at 62, 67 or 70; a metered run turns first-meeting notes into the plan sheet or writes the plan review from
> the page's numbers. A planning model, not financial advice. Built on the SkillSafe skill @anthropics/financial-plan
> (anthropics/financial-services, Apache-2.0).

https://retire-runway.skillsafe.ai/

## The input format
A plan sheet - Plan:, Year:, then People: ("Priya: age 52, retire at 62, plan to 95, salary 96,000"), Accounts:
("Priya 401(k): 410,000 pre-tax, adding 23,000 a year"; types pre-tax, roth, taxable, cash), Spending:, Retirement
spending:, Social Security: ("Tom: 2,400 a month at 67, claim at 62" - the benefit at full retirement age), Pensions:,
Mortgage:, Education: ("Maya: age 14, 32,000 a year for 4 years from 18, 529 balance 48,000, adding 3,000 a year"),
Events:, Assumptions: (inflation, raises, stocks %, stock return, bond return, tax rate), Insurance:, Estate: - or free
meeting notes, which the intake lane turns into a sheet.

## What is free
- The projection, year by year in today's dollars: pay with real raises, Social Security adjusted for the claiming age (5/9% a month early for 36 months, 5/12% beyond, 8% a year delayed), pensions (eroding unless inflation-indexed), a fixed mortgage, college costs paid from the 529 first, one-off events, withdrawals in the order cash, taxable, pre-tax (grossed up for tax), Roth, required minimum distributions from 73 (75 if born 1960 or later) on the Uniform Lifetime Table, a survivor spending 75% and keeping the larger Social Security benefit.
- A seeded 1,000-path Monte Carlo (stocks 17% and bonds 6% volatility), the probability of success, the 10th/50th/90th percentile portfolio, the sustainable retirement spending at 85% (bisection), and the stress test on the same paths: retire 2 years early, a 20% market drop in the first retirement year, spending +20%, living 5 years longer, long-term care at $100,000 a year for 3 years from 85; Social Security claimed at 62, 67 and 70; the monthly saving that would fund each college goal.
- Checks: the money running out, success under 85%, a first-year withdrawal rate over 4.5%, fragility to a market drop, optimistic returns, inflation out of range, working years spending more than they earn, heavy stocks near retirement, early claiming, underfunded education, dependants without life cover, no long-term-care plan, a will missing or older than five years, beneficiaries not reviewed in three.
- An SVG chart, the scenario, claiming, education and year-by-year tables, a printable plan summary, year-by-year CSV, findings CSV.

## The metered lanes
Two tasks on gpt-terra, metered per run, routed by `task`. Input string fields: `task` (intake or review), `plan` (the notes or the sheet), `focus` (optional) and `facts` (the page's work).
- `intake` replies with `summary`, `sheet` (people, accounts, spending, retirement spending, Social Security, pensions, mortgage, education, events, assumptions, insurance, estate), `questions`, `left_out` and `caveats`. The page requires every number on the sheet to come from the notes (or a month/year conversion), no page default copied in, and the sheet to read back with every person and account.
- `review` replies with `verdict` (on_track, borderline, off_track), `summary`, `drivers`, `scenarios`, `actions` (prioritised, each with an area, why and effect), `education`, `protection`, `questions`, `findings`, `prescan` and `caveats`. The page requires every number to be one it printed, the verdict to match its probability band, every stress test read, every error and warning answered, and no product, provider or promise.

## Links
- API tutorial: https://retire-runway.skillsafe.ai/api.html
- Source skill: https://skillsafe.ai/skill/@anthropics/financial-plan