Cost Sentinel

Zero-to-One Local FinOps VS Code Extension

1. Concept & Problem Discovery

• The Problem: Cloud data warehouse compute costs are heavily influenced by developer query behavior. However, cost feedback loops are reactive. Developers only realize a query was expensive after the AWS/GCP bill arrives.
• The User: Data Engineers, Analytics Engineers, and Backend Developers.
• The Hypothesis: Shifting FinOps entirely to the left—into the IDE—will prevent accidental compute spend by intercepting expensive queries before execution.

2. Product Requirements & Scoping (MVP)

• Core User Flow: A developer types a SQL query in VS Code. The IDE instantly estimates the cost and displays a red visual alert if it exceeds a pre-set budget.
• In Scope for V1: Real-time cost estimation, local configuration menu for table pricing, status bar UI injection.
• Out of Scope for V1: Native cloud API integration. Rationale: Network latency per keystroke ruins the developer experience, and requiring IAM credentials introduces massive security friction.

3. System Design & Trade-offs

To execute the MVP, I evaluated three architectural paths. I rejected Cloud APIs (too slow, privacy risk) and Regex Parsing (too fragile for nested queries).

Selected Architecture: Local AST (Abstract Syntax Tree) Parsing. Extremely fast, works completely offline, zero credential risk, and accurately identifies table structures.

4. Execution & Development

• Platform Choice: Built specifically for VS Code, capturing ~70% of the target developer market share.
• Tech Stack: TypeScript, Node.js, `node-sql-parser`, VS Code Extension API.
• Security Posture: 100% Air-Gapped. The tool reads text locally and never phones home, guaranteeing zero exposure of PII/PHI.

5. Go-To-Market & Distribution

• Marketplace Launch: Published directly to the Visual Studio Code Marketplace.
• Open Source Community: Repository open-sourced on GitHub to drive peer review and developer trust.