model-usage

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). At runtime, scripts/model_usage.py ingests and parses first-party outsider-authored content from CodexBar local cost logs by invoking codexbar cost (or reading --input JSON/stdin) and then uses fields like modelBreakdowns[].modelName/cost directly in its processing/output.