Checks GitHub for the latest release, downloads the new binary, and replaces the current executable when that path is writable. If your existing install is in a system directory, upgrade falls back to ~/.entropyfa/bin instead of prompting for sudo. The CLI also checks for updates in the background.
Quick Start
30-Second Demo
1. Discover what data is available:
entropyfa data coverage
2. Look up 2026 federal income tax brackets for single filers:
entropyfa data lookup --category tax --key federal_income_tax_brackets --year 2026 --filing-status single
Look up specific reference data by category, key, year, and applicable parameters. Lookup responses now include the value plus verification metadata and source URLs by default.
Usage
entropyfa data lookup [OPTIONS]
Flags
Flag
Description
Required
--category
Data category (tax, retirement, etc.)
Yes
--key
Data key within category
Yes
--filing-status
Filing status filter
When applicable
--year
Reference year (defaults to latest)
No
--lived-with-spouse-during-year
Required for IRMAA lookups when filing status is married_filing_separately
When applicable
Example
entropyfa data lookup --category tax --key federal_income_tax_brackets --year 2026 --filing-status single
The IRMAA contract distinguishes married-filing-separately taxpayers who lived with a spouse during the year from those who lived apart, because the official CMS and SSA rules use different bracket tables.
Use the global --result-hook-url flag to POST the same JSON envelope to a webhook endpoint.
compute federal-tax
Compute federal income + payroll taxes for a given filing status and income breakdown.
Usage
entropyfa compute federal-tax --json '<JSON>'
When to Use
When a user asks about their federal tax liability, effective tax rate, marginal rate, bracket mechanics, or payroll tax (FICA / SE) consequences.
Schema (What to Gather)
Field
Description
Required
filing_status
single, married_filing_jointly, married_filing_separately, head_of_household, qualifying_surviving_spouse
Yes
income.wages
W-2 wages
At least one income field
income.qualified_dividends
Qualified dividend income
No
income.long_term_capital_gains
Long-term capital gains
No
income.short_term_capital_gains
Short-term capital gains
No
income.taxable_ira_distributions
Taxable traditional IRA distributions
No
income.taxable_pensions
Taxable pension income
No
income.taxable_social_security
Taxable Social Security benefits
No
income.self_employment_income
Self-employment income
No
adjustments
HSA, IRA, student loan interest, other above-the-line adjustments
No
deductions.method
standard or itemized (defaults to standard)
No
deductions.itemized_amount
Aggregate itemized deductions when you already have the final Schedule A total
No
deductions.state_local_income_or_sales_tax
Elected state/local income tax or sales tax amount for Schedule A SALT
No
deductions.real_property_tax
State and local real property taxes
No
deductions.personal_property_tax
Deductible personal property taxes
No
deductions.other_itemized_deductions
Non-SALT itemized deductions to add after the SALT cap is applied
No
deductions.spouse_itemizes
Whether an MFS spouse itemizes, which can force itemized treatment
No
Use either deductions.itemized_amount or the detailed itemized fields, not both. deductions.state_local_income_or_sales_tax is the elected Schedule A income-or-sales-tax amount, not an automatic sales-tax-table calculation.
Run Monte Carlo or linear projection of portfolio balance over time. The CLI always returns the JSON envelope on stdout. The terminal dashboard is opt-in with --visual and renders to stderr in a real terminal.
Delivery: the CLI always writes the JSON envelope to stdout and, if --visual is set, renders the dashboard separately on stderr. If --result-hook-url is set, the same JSON envelope is also POSTed as application/json.
Reference data is compiled into the binary and returned with provenance. Use entropyfa data coverage to discover the surface area and entropyfa data lookup to retrieve both the value and its verification metadata.
Trust Model
authoritative means the entry is backed by primary official sources or reviewed maintainer metadata. corroborated means multiple approved secondary sources agree. derived and placeholder remain visible so consumers can make their own trust decision.
Pipeline Reviewed vs Metadata Backed
pipeline_reviewed: true means the entry has gone through a primary agent, verifier agent, human review, and reviewed-artifact apply step. In the current public release, all 30 shipped entries are in this state.
Category
Key
Description
Parameters
Years
tax
federal_income_tax_brackets
Federal income tax brackets by filing status
filing_status
2026
tax
federal_standard_deductions
Federal standard deduction amounts by filing status
filing_status
2026
tax
federal_salt_deduction_parameters
Federal SALT deduction cap, phaseout, and floor by filing status
filing_status
2026
tax
federal_capital_gains_brackets
Federal long-term capital gains tax brackets
filing_status
2026
tax
federal_net_investment_income_tax
Net investment income tax rate and thresholds
filing_status
2026
tax
federal_payroll_tax_parameters
Social Security, Medicare, and self-employment tax parameters
filing_status
2026
tax
federal_capital_loss_limit
Annual capital loss deduction limit
filing_status
2026
tax
federal_qbi_deduction
Qualified Business Income (Section 199A) deduction parameters
This table shows the embedded surface area. For live verification status, completeness, and source URLs, use entropyfa data lookup against the specific entry you care about.
Agent Integration
entropyfa is designed as a tool for AI agents doing financial planning. Here's how to integrate it, including the official OpenClaw skill.
The --schema Workflow
Agent calls entropyfa compute <command> --schema to learn what inputs are needed
Agent gathers required information from the user
Agent constructs JSON input and calls the command
Agent parses JSON output and reasons about results
Design Principles for Agents
--schema
on every command — agents read the schema to know what inputs to gather
data coverage
agents discover what reference data is available without hardcoding keys
JSON-in/JSON-out
structured I/O that agents parse natively
Deterministic
same input always produces the same output
No configuration
install and go, no API keys, no config files, and no outbound calls unless you explicitly use upgrade or result hooks
OpenClaw Quick Start
Install the local entropyfa CLI first, then install the official OpenClaw skill into your current workspace:
clawhub install entropyfa
The official skill is published as entropyFA Financial Planning. It provides verified financial planning data and blazing-fast, deterministic calculators for Monte Carlo projection, goal solving, Roth conversions, RMDs, income tax, estate tax, and pension analysis. Its recommended flow is:
Run entropyfa data coverage to discover available planning datasets.
Run entropyfa data lookup ... when the user needs official yearly values, trust status, or source URLs.
Run entropyfa compute <command> --schema when required inputs are missing.
Run entropyfa compute <command> --json '<JSON>' once the inputs are known.
Use compute projection --visual only when a human actually wants the terminal dashboard.
primary agent → verifier agent → human review → reviewed artifact
\
→ generated Rust data + metadata + snapshot
Key Properties
Local by default
Reference data and provenance are compiled into the binary; outbound calls are opt-in
Sub-millisecond
Pure computation, no I/O overhead
Single binary
No runtime dependencies
Versioned releases
Updated as reviewed reference data changes are promoted into the binary
Workspace Structure
Crate
Purpose
engine
Embedded reference data, provenance artifacts, and computation logic (usable as a Rust library)
cli
CLI that accepts JSON via --json flag, writes JSON envelopes to stdout, and can optionally POST result envelopes
Releases
entropyFA ships versioned releases as reviewed data updates are promoted into the binary. Each release bundles the current embedded dataset and provenance state at that version.