fully-review

A full codebase review partitions your repository into six subtrees and verifies each finding against cited evidence and a closed-findings log. This ensures broad, repeatable coverage that de-duplicates results to prevent regressions during large-scale refactors or release prep.

Finding triage during a repository audit works by checking each emitted finding via cited file:line evidence and grep call-site confirmation. Closed or speculative items are dropped, and remaining issues are recorded as standardized standalone files with severity, location, and suggested fixes.

De-duplicating static analysis findings requires cross-checking each emitted issue against a single-source closed-findings log. This verifies evidence via cited file:line references and grep call-site confirmation, dropping closed or speculative items to prevent duplicate reporting.

Yes, you can automate a code review workflow for periodic audits by initializing subagents with specific review rules and log-safety constraints. The workflow partitions the repository into six subtree batches, verifies findings, and consolidates results into an indexed task directory.

Code review findings are formatted as standardized per-finding files containing Severity, Location, Summary, Details, Impact, and Suggested fix. These standalone files are collected and consolidated into a tasks directory with an INDEX.md for easy review and indexing.

You should avoid six-way scope splitting when your repository lacks a clearly defined src/ directory, as this workflow partitions src/ into non-overlapping subtree batches. It is designed for broad pre-release or post-refactor coverage rather than targeted single-file reviews.