Portfolio: Claude skills backed by calculation code

What wealth-portfolio calculates

11 commands, each available in the CLI, as an MCP tool and as a JavaScript import. Every result carries the method it used, the questions it still needs answered and the next steps worth taking.

When to use it

• Evaluate portfolio asset allocation against targets and monitor drift.
• Run historical backtests (CAGR, Sharpe, Sortino, max drawdown) or forward Monte Carlo simulations.
• Calculate Value at Risk and expected shortfall.
• Find tax-loss harvesting opportunities across taxable lots.

Requests this pack answers:

• “is the portfolio off target”
• “do they need rebalancing”
• “what trades bring this back to target”
• “rebalance this portfolio”
• “how much could this portfolio lose”
• “what is the value at risk”
• “what is this portfolio made of”
• “what is the duration”
• “any losses to harvest”
• “tax loss harvesting opportunities”
• “how would this mix have done”
• “backtest a 60/40”
• “what happens if stagflation returns”
• “project this portfolio forward”
• “what is the basis on this sale”
• “which lots should we sell”
• “how did the manager perform”
• “time-weighted return”
• “what is the IRR on this fund”
• “money-weighted return”
• “is this Sharpe ratio significant”
• “could this backtest be luck”

What it does not do

Named here because leaving it out is how an agent ends up asserting something the code never computed.

Guidance only, no engine behind it

• Portfolio accounting, performance attribution and client reporting (Addepar, Orion, Black Diamond, APX, Geneva)
• Risk-model factor exposures (BarraOne, Axioma): the engine only averages factor scores you supply
• Portfolio optimization, tax-aware transitions and direct indexing

Limits to state with every result

• Tax-loss harvesting checks only purchases in the 30 days *before* the sale, across the lots supplied. Confirm separately that nothing substantially identical is bought in the 30 days after, including in IRA and spouse accounts. Replacement candidates track a different index, but whether a fund is "substantially identical" is a judgment for the advisor or tax professional. No replacement is suggested for total-bond funds.
• Backtests use approximate embedded annual returns and capture drawdowns at year-end only.
• Forward-simulation regime assumptions are illustrative, not forecasts; pass your firm's capital market assumptions for client work.
• VaR is parametric normal and only as good as the volatility you supply.

Run it

node bin/wealth-skills.js portfolio drift-monitor --current '{"equity":68,"fixedIncome":32}' --target '{"equity":60,"fixedIncome":40}' --value 1000000 --band 5
node bin/wealth-skills.js portfolio var --value 1000000 --vol 0.14 --confidence 0.99
node bin/wealth-skills.js portfolio tlh --lots '[{"id":"L1","symbol":"VOO","quantity":100,"purchasePrice":500,"currentPrice":420,"purchaseDate":"2025-02-03"}]'
node bin/wealth-skills.js quant backtest --weights '{"VTI":0.6,"BND":0.4}' --initial 100000
node bin/wealth-skills.js quant forward-test --weights '{"VTI":0.6,"BND":0.4}' --regime stagflation

The same commands are exposed as MCP tools, so Claude, Claude Code, Codex, Cursor and any other MCP client can call them directly. See the integration guide and the pack itself, skills/wealth-portfolio/SKILL.md.

Catalog coverage

91 resources from the US Wealth Management Capability Catalog map to this pack, 22 of them backed by engine code and the rest cited as guidance or reference.

Questions it answers

The other skill packs

• Compliance: Automated screen of advisory communications for promissory language and missing standard disclosures, FINRA 2111 suitability and Marketing Rule performance checks, investment adviser registration lookups in SEC IAPD compilation files, and guidance for supervisory review.
• CRM: Meeting transcript parsing into decisions and action items, and CRM task payload construction, plus guidance for meeting preparation.
• Execution: Pre-trade validation (buying power, oversell, short-term gain), order payload construction for Alpaca, Interactive Brokers and FIX 4.4, ISIN/CUSIP/FIGI/LEI check digits and T+1 settlement dates, for human-approved execution.
• Fund operations: Fund administration and wealth operations checks - position reconciliation against the custodian, NAV tie-out with level 3 and stale-price flags, LP capital statement roll-forwards, and month-end close tracking - plus guidance for accounting and regulatory reporting that the engines do not perform..
• Onboarding: Customer Identification Program field checks, application gaps by account type (CIP, beneficial ownership, FINRA 4512, 2090, 2111 and 2165), and onboarding milestone tracking.
• Planning: US retirement and tax planning calculations - tax year 2026 bracket headroom for sizing Roth conversions, required minimum distributions, retirement Monte Carlo with inflation-adjusted withdrawals, capital-loss netting and carryover, and net investment income tax - plus guidance for broader household planning that the engines do not model..
• Research: Valuation ratios from supplied fundamentals, DCF valuation with sensitivity, keyword risk-factor summaries of filing text, XBRL fact extraction, and live FRED and SEC EDGAR fetches, plus guidance for earnings and market research that the engines do not perform..
• UHNW: Concentrated-stock collar construction with constructive-sale review, and private equity multiple tracking (TVPI, DPI, RVPI), plus guidance for trust, philanthropic and exchange-fund planning that the engines do not model..