Labsco
MCP SERVER · OFFICIAL PROJECT

Cinderfi

by cinderfi

Tax-aware retirement planning for Canada and the US — projections, CPP and Social Security timing, Monte Carlo and withdrawal-order optimization.

Personal Finance & BudgetingOfficial source
Summary
Retirement questions answered with real tax math instead of rules of thumb.

The claiming-age and clawback questions are where general advice fails hardest, and these run the actual provincial and state rules rather than a national average. simulate_purchase is the one people do not expect: it prices a large expense in years of retirement rather than dollars, which is the form the decision is actually made in.

What it is

A retirement planning MCP server with real tax engines behind it: every Canadian province and US state, CPP, OAS, GIS and Social Security. It runs year-by-year projections, answers claiming-age questions with breakeven maths, and simulates what a purchase or a windfall does to the rest of the plan.

What you get
  • run_projection and run_full_analysis — a year-by-year retirement projection for one person, or the projection, score and insights in one call
  • calculate_score and generate_insights — a 0–100 readiness score, and personalized optimization recommendations
  • calculate_tax and estimate_benefit — income tax for any province or state, and CPP, OAS or Social Security at a chosen claiming age
  • calculate_cpp_timing and calculate_ss_timing — breakeven analysis across CPP ages 60 to 70, and Social Security 62 to 70 against full retirement age
  • calculate_gis and calculate_meltdown — GIS eligibility, clawback and provincial supplement, and an RRSP meltdown or Roth conversion strategy
  • simulate_purchase and simulate_windfall — what a one-time or recurring expense costs the plan, and where an inheritance or bonus is best allocated
  • compare_spending_strategies and map_financial_data — fixed, Guyton-Klinger and VPW withdrawal rules compared, and YNAB or bank exports converted into inputs
  • run_monte_carlo, run_backtest and compare_scenarios — probabilistic success with percentile bands, every historical window since 1871 on Shiller data, and one to four scenarios side by side
  • optimize_withdrawal_order and optimize_pension_split — the best claiming and withdrawal sequence, and pension income splitting to minimise combined tax
Requirements

A free API key from cinder.fi/mcp, then connect to https://mcp.cinder.fi/mcp over Streamable HTTP or SSE with Authorization: Bearer. Clients without remote support can bridge with npx -y mcp-remote. In Claude Code the plugin installs the server plus a skill that sequences the tools, and reads the key from CINDERFI_MCP_KEY. The free tier allows 5 calls a day and the 14 single-person tools; Monte Carlo, the optimizers, backtesting, scenario comparison and couples projections are on the paid tier.

Setup effort

One command plus a key — claude mcp add cinderfi --transport http https://mcp.cinder.fi/mcp -H "Authorization: Bearer YOUR_API_KEY", then supply credentials