FinPlan Skill
Auto-triggered skill that connects your AI assistant to the FinPlan MCP server. When you ask about financial projections, tax calculations, retirement planning, or any supported topic, the assistant automatically invokes the relevant MCP tools.
How It Works
Section titled “How It Works”- You ask a financial planning question in natural language
- The skill activates and determines which MCP tools to call
- Tools execute on the remote FinPlan server and return results
- The assistant interprets results and responds
No manual invocation needed — the skill triggers whenever your request matches a supported capability.
Supported Capabilities
Section titled “Supported Capabilities”The skill provides access to all MCP Tools. Their main areas:
| Category | Examples |
|---|---|
| Profile & State | Create/update your financial profile |
| Accounts | Add accounts, check limits, asset classes |
| Employer Match | 401(k) matching formulas, vesting |
| Tax | Federal income tax, capital gains, after-tax projections |
| Projections | Monte Carlo simulations, scenario comparison |
| Portfolio | Expected returns, volatility, glide paths |
| Goals | Define goals, track progress, required contributions |
| Budget | Income streams, expenses, cashflow |
| Scenarios | What-if plan changes, side-by-side comparison |
| Social Security | Benefits estimation, claiming strategies |
| Mortgage | Payments, amortization schedules |
| Liabilities | Credit card and loan payoff |
| RMDs | Required minimum distributions |
| Charts | Fan charts, account breakdowns, scenario comparisons |
| Snapshots | Point-in-time facts records, snapshot diffs |
| Account Sync | Link and reconcile accounts from an aggregator |
| Tool Search | Discover available tools by keyword |
Conventions
Section titled “Conventions”- Money is in cents (integers).
10000000= $100,000.00 - Rates are decimals.
0.07= 7% - Allocations are integers 0–100 that sum to 100
- Dates use
YYYY-MM-DDformat - Tools return file URLs for large datasets — the assistant uses
jqfor targeted queries rather than loading full files into context