<!-- markdownlint-disable-file MD041 -->
STARTER_CHARACTER = ⚡
Quick Review
Review work-in-progress and recent commits across 8 focus areas in parallel. No arguments. Auto-detects deltas.
Step 1: Detect Deltas
Compute what changed since the project root was last touched. Span committed, staged, unstaged, and untracked work.
bash
PROJECT_ROOT=$(git rev-parse --show-toplevel)
BASELINE=$(git log -1 --format=%H -- "$PROJECT_ROOT")
[ -z "$BASELINE" ] && BASELINE=$(git rev-list --max-parents=0 HEAD)
git log --oneline ${BASELINE}..HEAD
git diff ${BASELINE} --find-renames --stat
git diff ${BASELINE} --find-renames
git status --porcelain
git ls-files --others --exclude-standard
The diff has no
so it spans the baseline through the working tree, including staged and unstaged edits.
surfaces renames as
entries with both old and new paths so a pure rename does not collapse into an empty diff.
is the authoritative list of every modified, added, deleted, renamed, copied, and untracked entry; use it to confirm rename detection.
surfaces new untracked files. Files matching
are excluded.
Stop only if all four signals are empty:
empty,
empty,
empty, and
git log ${BASELINE}..HEAD
empty. Then report "No deltas to review" and stop.
Step 2: Read Untracked Files
For each file from
git ls-files --others --exclude-standard
, read the full file content. Untracked files have no diff representation.
Step 3: Dispatch 8 Review Agents in Parallel
Dispatch 8 parallel reviews, one per focus area. Use subagents if available, otherwise run sequentially:
| Agent | Subagent type |
|---|
| Architecture & Design | |
| Correctness & Bugs | |
| Operational Readiness | |
| Performance | |
| Code Quality | |
| Security | |
| SOLID Principles | |
| Testing | |
Each review prompt contains:
- The full diff output from Step 1
- The contents of any untracked files from Step 2
- The instruction: "Review the following changes for your focus area. Return findings as your response."
Do not pack repomix. Each agent has all the context needed in the prompt.
Step 4: Consolidate Findings
After all 8 agents return, produce a single consolidated report:
markdown
# Quick Review Report
## Summary
- Total findings: <N>
- High: <count>
- Medium: <count>
- Low: <count>
## Findings by Severity
### High
[All High findings from all 8 agents, ordered by file:line]
### Medium
[All Medium findings from all 8 agents, ordered by file:line]
### Low
[All Low findings from all 8 agents, ordered by file:line]
When to Use
This skill auto-loads when:
- A bug fix or feature implementation has just been completed
- Work is about to be committed or a PR is about to be opened
- A subagent-driven-development task has just passed spec compliance review
It is also slash-invokable for ad-hoc review of current work.