Methodology
How the study reads repository structure.
Biguá Analyzer uses public GitHub repository data to measure structural signals. The goal is not to label repositories as safe or unsafe, but to support careful interpretation of resilience, continuity, and dependency risk.
Repository Set
The repository set spans Go, JavaScript, Java, Python, and Rust projects across frameworks, infrastructure tools, programming languages, developer tooling, and other high-impact open-source categories.
Data Sources
Repository metadata
Project-level information used to contextualize size, age, activity, and adoption.
Commits, issues, PRs
Observable development behavior used for contribution, activity, and maintenance signals.
Releases and cadence
Maintenance rhythm and recent release behavior used as supporting structural context.
Structural Metrics
Contribution concentration
Gini coefficient and top contributor share measure inequality in contribution distribution across contributors.
Contributor redundancy
Bus factor at 50% and 75% estimates how many contributors account for half or three quarters of analyzed contribution activity.
Activity and maintenance
Release cadence, recent release cadence, commit volatility, and contributor inactivity windows help interpret project continuity.
Turnover
Developer turnover estimates churn across the analyzed contributor base and must be interpreted in project context.
Inferred AI influence
The AI layer combines reproducible repository-visible heuristics. It is not proof of AI authorship for individual commits.
Analysis confidence
The traffic-light classification describes signal quality and interpretive confidence, not repository goodness or security.
Fast Mode and Full Mode
Fast mode
Fast mode asks how the repository looks in recent observable behavior. It is useful for screening and prioritization across many dependencies.
Full-history mode
Full mode asks what the repository's structural profile looks like across its accessible history. It is the stronger validation step for sensitive decisions.
These modes are complementary and are not expected to always produce identical readings.
Turnover Definition
In this study, turnover means contributor-base churn in the analyzed window. It does not mean employee attrition, maintainer departure, or organizational staffing change.