cross-review

Automating GitHub pull request code review involves fetching diffs via the GitHub API to validate logic, security, style, and breaking changes. This classifies changes into levels L0–L3, assigning reviewers and merge permissions accordingly to enforce consistent cross-reviews and reduce merge risks.

Change level classification in risk assessment categorizes pull requests into L0–L3 tiers based on impact, such as docs, configs, CI, or constitution-level changes. This classification determines who reviews the code and what merge permissions are required to safely validate modifications.

To cross-review PRs created by another AI agent, you fetch the diff via the GitHub API and apply fixed role-based review rules that prevent self-approval. The system evaluates logic and security, then outputs an APPROVE, REQUEST_CHANGES, or COMMENT decision with justification.

Yes, role-based review rules can enforce a strict ban on developers approving their own pull requests. Fixed cross-review rules define exactly who reviews whose code, routing high-risk changes for triple verification to ensure no unauthorized merges occur.

The best way to enforce merge rules for constitution-level changes is to classify them as high-risk L3 modifications via cross-review. This triggers triple model verification and strictly validates logic, security, and breaking changes before granting the required merge permissions.

Yes, fetching diffs through the GitHub API using `gh api` is required for pull request validation. This allows the review process to programmatically access the exact code changes, validate logic and testing, and produce a final merge decision with justification.