codex-simplify

Codex Simplify

Run a post-implementation cleanup pass over recently changed files.

This skill is the Codex approximation of Claude Code's bundled /simplify behavior. It intentionally lives outside .claude/skills/ and uses the non-conflicting name codex-simplify so Claude Code keeps its native /simplify command.

Preserve the public contract:

• target recently changed files
• review for reuse, quality, and efficiency
• use three independent review passes
• aggregate findings
• apply fixes

Treat any user-supplied focus, such as memory efficiency, dead code, duplication, or readability, as extra weighting rather than a replacement for the core three review lenses.

When to use

Use this skill when:

• the code already exists
• the implementation or bug fix is already done
• the user wants a cleanup or refactoring pass rather than new feature work
• a workflow calls for a /simplify-style pass after review approval

Do not use this skill as a substitute for initial implementation, plan compliance review, or full validation.

Target file selection

Approximate "recently changed files" with this priority order:

1. Tracked unstaged changes in the current worktree
2. Tracked staged changes
3. Untracked files clearly related to the current task
4. If the tree is clean, files changed by the most recent commit

Use git to discover the set. Prefer file lists equivalent to:

git diff --name-only --diff-filter=ACMR
git diff --cached --name-only --diff-filter=ACMR
git ls-files --others --exclude-standard
git diff --name-only --diff-filter=ACMR HEAD~1..HEAD

Deduplicate the list. Exclude generated artifacts, lockfiles, build output, and unrelated files unless the user explicitly asks for them.

If there are no relevant changed files, stop and report that codex-simplify has nothing to do.

Review workflow

Run three independent review passes over the same target file set:

1. Reuse pass Look for duplication, repeated logic, avoidable branching, and places where a simpler shared helper or existing abstraction should be used.
2. Quality pass Look for confusing structure, dead code, weak naming, brittle conditionals, unnecessary complexity, and maintainability problems.
3. Efficiency pass Look for avoidable work, redundant allocations, repeated I/O, wasteful data transformations, and obvious performance footguns.