Codex loops I use.

Audit [project] against [product contract document]. First extract every concrete product promise, business rule, permission rule, background job expectation, user-facing workflow, non-goal, and quality bar from the contract into a requirement ledger. For each requirement, find current evidence in code, configuration, tests, docs, jobs, migrations, routes, UI, prompts, queues, and production-safe runtime state. Mark each requirement proved, weak, contradicted, unimplemented, N/A with reason, or blocked with missing evidence. Fix only confirmed high-impact mismatches one bounded slice at a time, preserving unrelated work, adding regression coverage where practical, and rerunning the affected ledger checks. Ask before touching production, secrets, user data, destructive operations, or changing public claims. Stop only when every requirement is proved or honestly classified and the ledger contains no silent gaps.

Continue optimizing the code for speed. After each significant change, measure page-load performance across every page under the same repeatable test conditions. Continue until every page loads in under 50 ms.

Run an SEO/GEO audit across crawlability, indexation, page intent, titles, internal links, structured data, source citations, and answer-first content. Rank the gaps by expected impact, fix the highest-leverage issue, then rerun the same crawl and target-query benchmark across search engines and AI answer engines. Repeat until no critical technical issues remain, every priority query maps to a clear answer-ready page, and the benchmark shows no high-impact gap left to fix.

Refactor until you are happy with the architecture. After each significant step, live-test the system, run autoreview, and commit. Track progress in /tmp/refactor-{projectname}.md.