reviewing-codelisted

# Code Review Produce findings, not edits. Review only the requested diff, PR, changed files, or file list. If scope or diff context is missing, ask one clarifying question. ## Read first Read `references/severity-rubric.md` before scoring or reporting findings. Load language references only for languages present in scope: - Go: `references/go.md` - Python: `references/python.md` - TypeScript: `references/typescript.md` - Web, HTML, CSS, JS, HTMX: `references/web.md` Unsupported language: use this skill and the severity rubric only; report reduced coverage. ## Review modes Default mode is standard unless the user asks otherwise. Quick: - Use for small, low-risk diffs. - Cover security and correctness only. - Review changed lines plus direct local context. Standard: - Use for normal reviews. - Cover security, correctness, and tests. - Review changed files plus direct callers or callees when needed to validate a finding. Deep: - Use when the user asks for deep, thorough, risk, or merge-safety review. - Cover all dimensions: security, correctness, tests, reliability, performance, maintainability, and docs. - Review changed files, relevant callers/callees, boundary inputs, and affected tests. Team: - Use only when the user asks for team, parallel, multi-agent, or multi-pass review. - Split by dimension or file group, then consolidate into one report. - Deduplicate by `file:line` plus claim. Keep the strongest severity only when evidence supports it. - Put unresolve