Technical debt is real and it has a cost, but most teams manage it by feel rather than measurement. "We know we have debt" is a common engineering team statement. "We have 340 hours of estimated remediation work, concentrated in the auth service and the data pipeline, 60% of which is AI-generated code that no human fully understands" is a different kind of statement — one that can drive prioritization, resourcing, and product roadmap decisions.
GitIntel's tech debt calculator runs five debt categories against your codebase: complexity debt (functions above cyclomatic complexity thresholds), coverage debt (modules below minimum test coverage), duplication debt (copy-paste code patterns), dependency debt (outdated packages, deprecated APIs), and AI comprehension debt (AI-generated code without adequate test coverage or documentation). Each category is quantified in estimated remediation hours using calibrated industry benchmarks.
AI comprehension debt is the newest and fastest-growing category. GitClear's 2026 analysis found AI tools pushed code churn up 41% — meaning the same code is being rewritten repeatedly rather than understood and improved. Each rewrite cycle adds to the debt rather than reducing it. Tracking this separately helps teams make the case for deliberate refactor investment before it compounds.
Run `gitintel debt` for a full report, or `gitintel debt --category ai` to focus specifically on AI-generated code risk.