Open source project / 2026
RepoPulse
An offline Git repository analytics and forensic CLI that inspects churn, author coupling, bus factor, and code velocity without external services.
Context
Engineering teams need visibility into code hotspot churn, knowledge silos, and bus-factor risks without sending proprietary source code to cloud analytics SaaS.
The problem
Legacy metrics like raw lines of code (LOC) fail to reveal structural codebase risk, developer concentration, or files that change dangerously often.
Approach
RepoPulse analyzes local git logs directly, calculating Pareto distributions of file churn, temporal coupling between modules, and ownership entropy.
System architecture
Zero-dependency Python leveraging git plumbing commands; computes co-change matrices and outputs both interactive terminal summaries and SARIF/JSON reports.
Key decisions
100% offline and privacy-first; requires only a local repository clone with zero telemetry, API keys, or network egress.
Outcome
Available on PyPI as git-repopulse. It flags fragile refactoring candidates, calculates bus factor across directories, and generates CI-ready audit reports.
Learnings
Codebase health is best understood through the velocity of change over time rather than static snapshot analysis alone.