Refactor (Lite)

Single-context refactor: audit → plan → approve → implement, all in this Opus session with no sub-agent fan-out, no intermediate task files. Behavior preservation is the primary success criterion. Verifies tests still pass, optionally commits. Review is a separate pass — offered at the end via /code-review.

Use this by default. Reach for the heavier /refactor only when the cleanup spans many independent files that can be parallelized, or genuinely exceeds one context.

Input

$ARGUMENTS — a file, directory, or description of what to improve.

Parse flags: if $ARGUMENTS contains --cross-review, set CROSS_REVIEW=true and strip it (else CROSS_REVIEW=false) — adds an opt-in GPT/Codex second opinion on the diff at the review handoff (Step 7). The remainder is the target.

Step 1: Audit

• Read the target(s). Explore surrounding code and call sites read-only.
• Identify concrete issues with file:line specifics: duplication, complexity, dead code, poor naming, missing abstractions.
• Check the target's test coverage. If coverage is thin, flag it — a safety net (Step 3) may be warranted.

Step 2: Plan

Compose a concise plan:

• What's wrong now — the smells worth fixing.
• The incremental changes, in safe order (smallest behavior-preserving steps first).
• Whether to write a test safety net first (recommend it when coverage is thin and the code is non-trivial).
• What will prove behavior is preserved — which tests.

Step 2a: Plan convergence — Codex adversarial cross-check (always-on)

Before presenting the plan for approval, run a read-only, fail-soft second opinion on it from a decorrelated model (GPT via the Codex CLI). It never blocks — if Codex is unavailable it logs SKIPPED and you present the plan as-is.

1. Resolve the helper: CODEX_REVIEW="$HOME/.claude/scripts/codex-review.sh"; [ -f "$CODEX_REVIEW" ] || CODEX_REVIEW="scripts/codex-review.sh". If neither exists, log "Plan convergence skipped — codex-review.sh not installed" and present the plan as-is.

2. Resolve a branch-scoped artifact dir:

   RAW_BRANCH=$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo "")
   SANITIZED=$(echo "${RAW_BRANCH:-nobranch}" | tr '/' '-' | tr -cs 'A-Za-z0-9._-' '-' | sed 's/^-*//; s/-*$//')
   CODEX_DIR="tasks/${SANITIZED:-nobranch}"; mkdir -p "$CODEX_DIR"
   

3. Write $CODEX_DIR/codex-plan-prompt.md = the plan text followed by this instruction verbatim:

   You are an adversarial refactoring reviewer with different training than the planner. Do NOT rewrite the plan or produce code. Challenge it specifically on: whether each step preserves observable behavior, the soundness of the decomposition and step ordering, and regression risk per step. Also flag missing safety nets and simpler/safer sequencing. Rank every finding as BLOCKER, MAJOR, or MINOR. If the plan is sound, say so explicitly. Be concise.

4. Run bash "$CODEX_REVIEW" "$CODEX_DIR/codex-plan-review.md" "$CODEX_DIR/codex-plan-prompt.md". Read the result from $CODEX_DIR/codex-plan-review.md (the helper writes Codex's output there; it is either a single SKIPPED: <reason> line or free-text findings tagged BLOCKER / MAJOR / MINOR):

   • SKIPPED: → note it; present the plan as-is.
   • BLOCKER/MAJOR present → revise the plan to address the valid findings before presenting it; tell the user what changed (and any findings you deliberately reject, with why).
   • MINOR only / none → present the plan, noting any MINOR items as advisory.

   This is a single inline pass — converge once, then present. The user may always override and proceed with the plan as-is.

Then present the (possibly revised) plan and ask via AskUserQuestion: "Proceed?" → Proceed / Revise (loop) / Cancel. Never edit before approval.

Step 3: Safety net (optional)

If the plan calls for it, write characterization tests for the current behavior first and confirm they pass against the unmodified code. This is your regression guard.

Step 4: Implement

Apply the refactor incrementally in this session. Preserve public APIs unless a breaking change was explicitly approved. No feature changes — refactor only. Keep context warm; no task files, no sub-agents.

Step 5: Verify

• Run the build/lint command. Skip if none.
• Run the full test suite. All previously-passing tests must still pass — that is the success criterion.
• If anything regresses: report it and ask whether to fix, proceed, or revert (git restore / git checkout --).

Step 6: Commit (optional)

Ask: "Commit these changes?"

• No → Step 7.
• Yes: if on main/master, branch to refactor/<3-5-word-slug> first; git add -A && git commit -m "refactor: <summary>"; then optionally push + open a PR (git push -u origin <branch> + gh pr create, or print the manual command).

Step 7: Review handoff

If CROSS_REVIEW=true, first run a cross-model second opinion on the diff (read-only, fail-soft — GPT via the Codex CLI; never blocks):

1. Resolve the helper: CODEX_REVIEW="$HOME/.claude/scripts/codex-review.sh"; [ -f "$CODEX_REVIEW" ] || CODEX_REVIEW="scripts/codex-review.sh". If neither exists, log "Cross-review skipped — codex-review.sh not installed" and continue to the handoff.
2. Resolve a branch-scoped artifact dir:

   RAW_BRANCH=$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo "")
   SANITIZED=$(echo "${RAW_BRANCH:-nobranch}" | tr '/' '-' | tr -cs 'A-Za-z0-9._-' '-' | sed 's/^-*//; s/-*$//')
   CODEX_DIR="tasks/${SANITIZED:-nobranch}"; mkdir -p "$CODEX_DIR"
   

3. Capture the diff into $CODEX_DIR/codex-diff-prompt.md. Uncommitted work shows up in git diff; if you committed in Step 6 you're on a feature branch, so also include the branch diff against the base. Resolve the base branch dynamically, then write the diff:

   BASE=$(git symbolic-ref refs/remotes/origin/HEAD 2>/dev/null | sed 's@.*/@@'); BASE="${BASE:-main}"
   { git diff; git diff "$BASE"...HEAD 2>/dev/null; } > "$CODEX_DIR/codex-diff-prompt.md"
   

   If $CODEX_DIR/codex-diff-prompt.md is empty, log "Cross-review skipped — empty diff" and continue to the handoff.
4. Append this instruction verbatim to $CODEX_DIR/codex-diff-prompt.md (after the diff text):

   You are a second reviewer of a refactor, with different training than the primary reviewer. Review ONLY the diff below. A refactor must preserve observable behavior — focus on any change in observable behavior, public API/contract changes, lost edge cases, and regression risk. Do NOT suggest cosmetic or stylistic changes. Rank each finding BLOCKER, MAJOR, or MINOR with a file:line reference. If behavior is preserved, say so explicitly. Be concise.

5. Run bash "$CODEX_REVIEW" "$CODEX_DIR/codex-diff-review.md" "$CODEX_DIR/codex-diff-prompt.md". Read the result from $CODEX_DIR/codex-diff-review.md; unless it starts with SKIPPED:, surface the GPT findings to the user (BLOCKER/MAJOR prominently). Advisory only — it does not block.

Then end with AskUserQuestion: "Run an independent /code-review pass?"

• "Yes" → invoke the code-review skill, scoped to the diff.
• "No" → print "Done. Run /code-review when you want an independent pass." and stop.

Rules

• Behavior preservation is the primary success criterion — a refactor that breaks tests is a failure.
• One context, no fan-out. If you want to spawn implementers, the job is big enough for /refactor.
• Never skip Step 2 approval before editing.