From wtf
Reviews recent git diffs and commits with brutal honesty before PRs, spotting 2am logic flaws, copy-paste artifacts, debug leftovers, hacks, and poor naming.
How this skill is triggered — by the user, by Claude, or both
Slash command
/wtf:was-i-thinkingThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
This is NOT a general code review. Review the user's recent *changes* — their diffs, their commits — not the codebase as a whole.
This is NOT a general code review. Review the user's recent changes — their diffs, their commits — not the codebase as a whole.
The difference from wtf:is-this: that skill reviews code quality and refactors. This skill reviews the decisions made recently — the kind of stuff a linter can't catch and only a human reviewer would notice.
Gather the changes. If $ARGUMENTS specify a target (PR number, commit range, branch, file path), go there directly — use whatever tools are appropriate (e.g., gh pr diff, git diff branch..branch, git show). If no arguments are given, fall back to local state: collect both unstaged and staged diffs, or recent commits if nothing is in flight.
Review like a human, not a linter. Look for the things automated tools miss:
Deliver the review:
**Overall:** [One brutally honest sentence]
**The Good:** [What's solid — give credit where due]
**The Suspicious:**
- [File:line] — [What looks off and why]
**The "Were You Asleep?":**
- [Obvious issues — typos, debug artifacts, copy-paste errors]
**Verdict:** [Ship it / Fix these first / Sleep on it]
wtf:is-this for automated fixes.npx claudepluginhub pacaplan/wtf --plugin wtfCode review with semantic diffs, expert routing by file type, and auto-task creation for critical issues. Works on staged changes, files, or PRs.
Reviews git changes with evidence-backed findings and risk-aware verdicts. Supports commit, range, file-scoped analysis, impact assessment, breaking-change detection.
Provides an independent code review by invoking a different AI model with zero conversation context. Useful for second opinions on code, commits, plans, or files.