Self-hosted alternatives to multi-reviewer-patterns

M

multi-reviewer-patterns

Coordinate parallel code reviews across multiple quality dimensions

multi-reviewer-patterns is an agent skill from the wshobson/agents collection for coordinating several code reviews that run in parallel, each looking at a different quality dimension of the same change. Instead of one reviewer trying to check everything, the work is divided so that each reviewer covers one dimension, such as correctness, security or performance, and the results are then combined. The skill covers deciding which review dimensions to assign to which reviewer, deduplicating findings when two reviewers report the same problem, calibrating severity ratings so that a finding means the same thing whichever reviewer raised it, and producing a single consolidated report for the author. Calibration and deduplication are the parts that make parallel reviews useful rather than noisy. The skill applies when organizing a review that spans several dimensions, deciding which dimensions a change needs, merging findings from several reviewers, keeping severity labels consistent, and writing the consolidated review. It is meant for teams and agent setups that review large or risky changes and want thorough coverage without one long, unfocused review. The output still goes to a human author, who decides which findings to act on. The skill is installed with the skills CLI from the wshobson/agents repository. Its skills.sh page showed about 9.5K installs and passing results from three security audits at the time of the scan.