ReWeaver AI Launches Deterministic Codebase DriftDetector

Summary

ReWeaver AI has introduced DriftDetector, a repository scanner designed to identify code that may pass review while silently diverging from intended standards or production requirements. Users can submit public or private GitHub repositories, and the tool evaluates nine production-readiness dimensions. It produces a standardized Production Drift Ratio from 0 to 1, grouped into severity bands and weighted by estimated fix hours, so different repositories or scans of the same repository can be compared. Findings are tied to specific files and lines, while commit history is scored to show when a gap appeared and whether it is widening or closing. Reports can be viewed online or downloaded as JSON, PDF, or Markdown. DriftDetector is rule-based and deterministic rather than an LLM-powered analyzer: repeated scans of the same repository are intended to return the same score, without token usage, prompts, or per-scan model costs. ReWeaver says repository contents stream directly from GitHub into temporary scanner memory and are not written to disk; historical charting uses commit metadata rather than file contents. The service says it retains no private-scan clone, scores, repository name, or report, while an encrypted GitHub token remains stored until the user disconnects or revokes access. The product also estimates hours saved and distinguishes potential future auto-fixes from issues requiring human judgment.