AI_Cost_Governor

Overview

Minimal Python control layer for interacting with AI models with explicit cost tracking and enforced secret hygiene.

Built to demonstrate:

• cost-aware AI usage
• deterministic execution
• practical DevSecOps guardrails

---

Key Capabilities

• Per-call cost visibility Surfaces tokens, cost, and running totals

• Secret-safe by default .env.local isolation + commit blocking via detect-secrets

• Pre-commit enforcement Prevents accidental leakage of API keys and sensitive data

• Structured runtime outputs Logs cost data for inspection and extension

---

Structure

src/        # application logic
config/     # local secrets (.env.local, not committed)
data/       # runtime logs
audit/      # local scan outputs (ignored)
docs/       # documentation

---

Security Model

• Secrets never enter version control
• Audit artefacts are quarantined
• Pre-commit scanning blocks unsafe commits

This repo is designed to make accidental leakage structurally difficult, not just discouraged.

---

Run

python src/main.py "your prompt"

---

Intent

A minimal, extensible foundation for:

• AI cost governance
• safe API usage patterns
• lightweight DevSecOps integration

---

Notes

• Requires local .env.local with API key
• Rotate any exposed keys immediately
• Designed for clarity over completeness