Parallel PR Reviewer FAQs

Question 1

Can I review specific files instead of the whole PR?

Accepted Answer

Yes, you can use the `--files <paths>` option to limit the review to specific directories or files, or use the `--diff-only` flag to focus strictly on changed lines.

Question 2

Does this skill post comments directly to my GitHub Pull Request?

Accepted Answer

No, this skill is designed to be non-intrusive. It saves all findings, validation reports, and summaries to a local `.reviews/` directory in markdown format, allowing you to review them before taking action.

Question 3

Which review tools are required for this skill to work?

Accepted Answer

This skill requires both the `code-review:code-review` plugin (official) and the `pr-review-toolkit:review-pr` plugin to be installed in your Claude Code environment.

Question 4

Is it possible to adjust the sensitivity of the review?

Accepted Answer

Absolutely. You can use the `--confidence` parameter to set a minimum threshold (default is 70) for findings to be included in the final aggregated summary.

Question 5

How does the validation phase work?

Accepted Answer

After the initial reviews are complete, the skill launches secondary 'Validator' agents that cross-reference findings, check for false positives, and assign confidence scores to ensure you only see the most relevant issues.

Parallel PR Reviewer

主要功能

使用场景

Parallel PR Reviewer

主要功能

使用场景