review

Coordinate structured code review across multiple reviewer personas for git diffs.

1|Updated Oct 17, 2025
One-click install
npx skills add https://github.com/softwaresalt/csv-managed --skill review-softwaresalt
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/softwaresalt/csv-managed/tree/main/.github/skills/review
Command: npx skills add https://github.com/softwaresalt/csv-managed --skill review-softwaresalt

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill reduces the risk of shipping bugs and regressions by coordinating structured code review across multiple reviewer personas and enforcing severity-based action routing.

Core Features & Use Cases

  • Tiered persona review: Runs always-on constitutional and Rust correctness review plus conditional domain-focused security/architecture/concurrency review based on the diff.
  • Structured, merged findings: Collects persona results, deduplicates matching issues, escalates disagreement conservatively, and assigns final severity (P0–P3).
  • Mode-controlled outcomes: Supports interactive review with user decisions, deterministic safe_auto autofix application, or report-only output suitable for build loops.

Quick Start

Run a report-only review for a branch named feature/add-indexing changeset to get structured findings without modifying files.

Frequently Asked Questions about review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an automated code review on a git diff before opening a pull request?▼

You can run a pre-PR code review by routing a git diff to multiple reviewer personas and merging deduplicated, severity-tagged findings. This Skill supports pre-PR, build-gate, and standalone diff reviews with confidence-gated recommendations.

What is severity gating in a structured code review process?▼

Severity gating in a structured code review assigns final severity levels from P0 to P3 to deduplicated findings. It enforces action routing and escalates persona disagreements conservatively to ensure confidence-gated recommendations.

Can I use persona routing to apply different code review checks to a standalone diff?▼

Yes, you can use persona routing to apply different code review checks to a standalone diff. It runs always-on constitutional and Rust correctness reviews while conditionally triggering security, architecture, and concurrency reviews based on path and content signals.

Does the code review support autofix and report-only modes for build loops?▼

Yes, the code review supports interactive, deterministic safe_auto autofix, and report-only modes. Report-only mode generates structured findings without modifying files, making it suitable for continuous integration build loops.

How do you deduplicate code review findings from multiple reviewer personas?▼

You deduplicate code review findings from multiple reviewer personas by collecting all results and merging matching issues. The system escalates disagreements conservatively and assigns a final severity to produce a unified, structured report.