ce-code-review

You can automate code review with fix suggestions by running the tool using the 'mode:autofix' argument. This triggers confidence-gated automated fixes that apply only policy-allowed safe_auto corrections to your code changes.

Confidence-gated automated code review evaluates code quality and security by applying tiered reviewer personas. It only applies fixes deemed safe by policy thresholds, then merges and deduplicates findings into a single structured report.

You get structured findings from an automated code review by utilizing dynamic reviewer personas and parallel sub-agents. These components analyze your code changes and return structured JSON findings detailing quality, security, and compliance issues.

Yes, you can use automated code review during iterative implementation. It is designed to provide structured feedback on code changes after completing a task or before creating a PR, ensuring code quality and compliance with project standards.

Yes, the code review process supports generating a single comprehensive report through its merge and dedup pipeline. This pipeline consolidates all findings from parallel sub-agents into one unified report for easier analysis.

The main limitation of confidence-gated fixes is that they only apply policy-allowed 'safe_auto' fixes automatically. Complex compliance or security issues requiring manual intervention are flagged in the structured findings but not auto-corrected.