Data Profiler

Description

# @aleexpe03/n8n-nodes-finance-data-tools

Community node package for finance-oriented data quality, deterministic finance calculations, reporting, predictive analytics, and AI Agent wrappers in n8n.

This package is currently prepared as a self-hosted MVP. It is not published to npm from this repository state, and official n8n verification is not claimed.

Implemented nodes

| Node | Purpose | Main operations |
| — | — | — |
| Data Profiler | Profiles tabular JSON rows before finance preparation. | profileDataset |
| Data Cleaner | Cleans finance-oriented tabular rows. | cleanDataset |
| Data Normalizer | Maps and normalizes finance columns. | normalizeDataset |
| Math & Statistics | Runs allowlisted statistics and finance-safe math. | Summary statistics, percentile, correlation, growth, CAGR, z-score, IQR outliers, grouped aggregates |
| Financial Ratios | Calculates closed financial ratios from one aggregated finance object. | Profitability, liquidity, leverage, returns, growth, cash-flow, efficiency, unit economics |
| Accounting Validator | Validates journals and can generate a minimal trial balance. | Validate journal entries, validate and build trial balance, build trial balance |
| Financial Report Builder | Builds deterministic financial report JSON. | Executive summary, P&L, balance sheet, cash summary, KPI table, dashboard JSON, AI Agent report |
| Predictive Analytics | Runs explainable directional forecasting. | Moving average, CAGR forecast, trend forecast, simple linear regression |
| AI Finance Tool | Controlled facade for AI Agents. | Data, finance, accounting, report, and opt-in predictive operations |

Documentation

• Self-hosted installation and local development
• AI Agent wrapper guide
• Output contract
• Security and limitations
• Testing strategy
• Release checklist

Example workflow templates

Editable workflow templates live in examples/workflows/:

• data-quality-wrapper.workflow.json
• financial-ratios-wrapper.workflow.json
• accounting-validator-wrapper.workflow.json
• ai-finance-tool-wrapper.workflow.json

The templates are intentionally marked as editable. If your n8n instance reports an unknown custom-node type during import, install/load this package first and confirm the exact node type shown by your n8n version.

Self-hosted status

The MVP target is self-hosted n8n. Current documentation does not promise n8n Cloud availability or official verified-community-node status.

Important verification note: the package currently uses runtime dependencies (date-fns, decimal.js, simple-statistics, and zod). n8n’s current verification guidance expects verified packages to avoid external runtime dependencies, so official verification requires a dependency strategy review before submission.

Local development commands

npm test
npm run lint
npm run dev

npm run dev is the local n8n development path provided by the n8n node tooling. npm run build exists in package.json, but it is intentionally not part of this phase’s executed validation.

Standard output envelope

All nodes return a shared envelope:

{
  "success": true,
  "operation": "profileDataset",
  "data": {},
  "metadata": {
    "generatedAt": "2026-05-09T00:00:00.000Z",
    "durationMs": 0,
    "rowCount": 0,
    "columnCount": 0
  },
  "warnings": [],
  "errors": [],
  "auditTrail": []
}

See Output contract for field-level semantics and predictive-output details.

Architecture

The n8n nodes are thin adapters. Domain logic lives in pure TypeScript services under:

• domain/data/profiling/
• domain/data/cleaning/
• domain/data/normalization/
• domain/math/statistics/
• domain/finance/ratios/
• domain/finance/reports/
• domain/accounting/
• domain/ml/forecasting/
• domain/ai/financeTool/

Shared output, audit, error, and type helpers live in shared/.

Safety model

• No Python runtime.
• No external APIs or microservices.
• No filesystem or environment-variable access in domain logic.
• No eval, formula interpreter, or arbitrary user-code execution.
• Operations are allowlisted and deterministic.
• Predictive output is directional, not a guarantee, and must surface warnings.

See Security and limitations for the full model.