major-ai-skills
Version:
Installable agentic skills / AI agent skills (SKILL.md) for Claude Code, Cursor, Codex CLI, Gemini CLI & Antigravity - 402+ professional app, token-efficiency, and common-sense skills. SEO/GEO ready.
110 lines (85 loc) • 5.27 kB
Markdown
name: keep-sensitive-financials-private
description: "Prepare redacted or synthetic financial inputs for AI-assisted analysis while keeping confidential source values private."
category: common-sense
risk: safe
source: self
source_type: self
date_added: "2026-08-26"
tags: ["data-privacy", "anonymization", "financial-safety", "security", "confidentiality", "prompt-engineering"]
tools: ["claude", "cursor", "gemini", "codex", "chatgpt"]
# Keep Sensitive Financials Private (Data Anonymization) (AI Skill)
## Overview
Pasting confidential company profit & loss (P&L) statements, employee salary sheets, bank account numbers, or secret cap tables into public AI models poses serious corporate security, compliance (GDPR/SOC2), and competitive risks.
The **Data Anonymization & Synthetic Scaling Protocol** enables you to leverage AI's advanced mathematical and financial reasoning capabilities while keeping 100% of your real confidential financial numbers and customer identities secret.
## The Synthetic Scaling Architecture
```
┌─────────────────────────────────────────────────────────────┐
│ Synthetic Anonymization Flow │
│ │
│ [ REAL CONFIDENTIAL DATA ]: │
│ • Client: Acme Bank │
│ • Actual ARR: $1,420,000 │
│ • Burn Rate: $85,000/mo │
│ │ │
│ ▼ │
│ [ LOCAL SANITIZATION (Apply 2.5x Multiplier + Alias) ]: │
│ • Client: "Enterprise Client Alpha" │
│ • Synthetic ARR: $3,550,000 │
│ • Synthetic Burn: $212,500/mo │
│ │ │
│ ▼ │
│ [ PROMPT AI FOR FORMULAS & ANALYSIS (Zero Privacy Risk) ] │
│ • AI calculates runway ratios and margin percentages │
│ • Ratios are mathematically identical to the real numbers! │
└─────────────────────────────────────────────────────────────┘
```
## The 3 Rules of Safe Financial Prompting
### 1. The Proportional Scaling Trick
Multiply or divide all your raw numbers by a fixed constant (e.g. $1.5\times$ or $10\times$).
Because financial metrics (Gross Margin %, Runway in Months, CAC-to-LTV ratio, Year-over-Year Growth %) are proportional, the AI's financial analysis will be **100% mathematically valid**, while your real confidential balance remains completely hidden.
### 2. Entity Aliasing (Search & Replace)
Never include real company names or individual personnel:
- Replace *"JPMorgan Chase"* with *"Tier-1 Enterprise Bank"*.
- Replace *"Sarah Connor ($180k/yr)"* with *"Lead Engineer A ($X/yr)"*.
### 3. Strip Account Identifiers
Always purge credit card numbers, routing numbers, Tax IDs (EIN/SSN), and wire transfer codes.
## Master Anonymized Financial Prompt Templates
### Pattern 1: The Scaled Runway & Unit Economics Analysis
```markdown
Analyze the unit economics of this SaaS company based on anonymized, normalized metrics:
Financial Snapshot (Indexed/Normalized Units):
- Monthly Recurring Revenue (MRR): 100 units
- Cost of Goods Sold (Hosting + Support): 22 units
- Sales & Marketing Spend: 35 units
- R&D / Engineering Payroll: 40 units
- General & Administrative: 15 units
- Net Monthly Burn: 12 units
- Cash in Bank: 180 units
Task:
1. Calculate the Gross Margin percentage.
2. Calculate the exact cash runway in months.
3. Provide 3 specific recommendations to reach break-even cash flow.
```
## Real-World Comparison
### Scenario: Reviewing an Acquisition Term Sheet
#### Risky Prompt (Confidential Leak)
> ❌ *"Here is the confidential term sheet from Microsoft offering $18.5M for my startup WidgetCo with a 4-year earnout for founders Alice and Bob..."*
> *(Violates NDA and exposes deal valuation to third-party model training).*
#### Anonymized Safe Prompt
> ✅ *"Review this anonymized acquisition term sheet for a B2B SaaS company:
> - Offer: $X Cash at closing + 25% Earnout tied to Year 2 revenue targets.
> - Founder Retention: 3-year non-compete and 2-year vesting clawback.
>
> What are the 3 biggest contractual and tax traps founders face in this structure?"*
## Privacy Checklist Before Hitting Send
- [ ] Are all company, customer, and employee names converted to aliases?
- [ ] Are absolute dollar amounts scaled or converted to percentages/ratios?
- [ ] Have all bank account numbers, tax IDs, and addresses been deleted?
- [ ] Does this prompt comply with your organization's corporate data security policy?