safe-mass-index-core

Bounded, metadata-only indexing prevents rg churn by enforcing strict budgets for files, time, and read bytes. This approach limits CPU thrash while enabling deterministic discovery by path, extension, and language for large codebases.

Bounded metadata-only indexing schedules repository analysis using strict limits on files, time, and read bytes. It captures path, extension, language, scope, and freshness data without reading full file contents, preventing CPU thrash and rg churn.

Yes, you can perform incremental or full builds on very large Python repositories. The indexing workflow supports deterministic discovery and outputs manifest, state, and run artifacts to ensure reproducible results across incremental updates.

To get reproducible results when analyzing large codebases, use deterministic bounded indexing that outputs manifest, state, and run artifacts. These artifacts capture the exact indexing state and budgets applied, ensuring consistent builds every time.

The primary limitation of bounded indexing is that it is metadata-only, meaning it does not index full file contents. It relies on strict budgets for files, time, and read bytes, which may intentionally skip files if limits are exceeded during large repo processing.