pr-review-check-suggestion

You can validate PR review findings by running a deterministic pre-output verification workflow that checks subagent results against current best practices. This Skill intercepts findings, uses optional web search to confirm issues and fixes, and assigns confidence levels to ensure accurate code review suggestions.

Confidence calibration in code review adjusts the certainty of findings when external knowledge is uncertain or unverified. You need it when automated PR review subagents rely on external information that may be outdated, ensuring reviewers only receive findings with appropriate confidence levels.

Yes, web search can verify code review issues when available by checking PR findings against current best practices. This Skill applies web search results during validation to confirm that identified issues and suggested fixes align with up-to-date standards before output.

Preventing prompt injection and data leakage in PR review subagents requires a deterministic workflow that enforces safety by design. This Skill validates findings through a controlled process that blocks prompt-injection vectors and prevents data leakage while surfacing accurate code review results.

Yes, this PR review validation works without external dependencies or components. The Skill operates as a standalone pre-output validation layer with no required dependencies, using an optional web search integration to verify findings and calibrate confidence when available.