Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Analyze code changes between two local Git branches and perform a comprehensive code review. Use ONLY for local git branch comparisons. Do NOT use when a GitLab or GitHub MR/PR URL or MR ID is provided — use gitlab-cli or github-cli skill instead.
.claude/skills/abdullahkhawer-code-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 190% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 202% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -38% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -42% | 0% |
Run this skill when asked to review code, compare branches, or perform a code review between a source and target branch using local git. Do NOT use this skill if the user provides a GitLab or GitHub MR/PR URL or MR ID — use the gitlab-cli or github-cli skill instead.
Ask the user the following two questions together if not already provided:
Kindly let me know:
1. Source Branch: What is the name of the source branch (the branch with new changes)?
2. Target Branch: What is the name of the target branch (usually main/master)?If the user has already provided either value upfront, skip the corresponding question.
bashgit diff --name-only <TARGET_BRANCH> <SOURCE_BRANCH>
If the output is empty, report that there are no differences between the two branches and stop.
For each file returned in Step 2, read the full diff:
bashgit diff <TARGET_BRANCH> <SOURCE_BRANCH> -- <FILE_PATH>
Perform a comprehensive code review across all changed files covering:
For each issue found:
Only include code snippets that actually require changes — do not repeat code that looks fine.
If no issues are found, state clearly that the code looks good and summarize what was reviewed.
git diff <TARGET_BRANCH> <SOURCE_BRANCH> — never checkout branches or modify the working tree.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,941 | 12,297 | -12% | 1 | 1 | 0% | 1,707 | 1,976 | +16% | 0 | 0 | — |
case-02 | fail→fail | 4,143 | 5,245 | +27% | 1 | 1 | 0% | 764 | 835 | +9% | 0 | 0 | — |
case-03 | fail→fail | 5,995 | 4,203 | -30% | 1 | 1 | 0% | 1,164 | 760 | -35% | 0 | 0 | — |
case-04 | fail→fail | 10,262 | 9,908 | -3% | 1 | 1 | 0% | 869 | 1,784 | +105% | 0 | 0 | — |
case-05 | fail→pass | 2,498 | 4,464 | +79% | 1 | 1 | 0% | 421 | 1,223 | +190% | 0 | 0 | — |
case-06 | fail→pass | 3,793 | 3,797 | +0% | 1 | 1 | 0% | 380 | 1,148 | +202% | 0 | 0 | — |
case-07 | fail→pass | 6,925 | 1,332 | -81% | 1 | 1 | 0% | 1,214 | 755 | -38% | 0 | 0 | — |
case-08 | fail→pass | 7,414 | 1,505 | -80% | 1 | 1 | 0% | 1,413 | 824 | -42% | 0 | 0 | — |
case-09 | fail→pass | 7,688 | 1,717 | -78% | 1 | 1 | 0% | 1,410 | 895 | -37% | 0 | 0 | — |
case-10 | fail→fail | 5,579 | 4,029 | -28% | 1 | 1 | 0% | 954 | 699 | -27% | 0 | 0 | — |
case-11 | fail→pass | 9,602 | 3,249 | -66% | 1 | 1 | 0% | 1,796 | 1,192 | -34% | 0 | 0 | — |
case-12 | pass→pass | 5,199 | 1,958 | -62% | 1 | 1 | 0% | 909 | 863 | -5% | 0 | 0 | — |
case-13 | pass→pass | 7,859 | 3,163 | -60% | 1 | 1 | 0% | 1,527 | 1,202 | -21% | 0 | 0 | — |
case-14 | pass→pass | 7,757 | 2,305 | -70% | 1 | 1 | 0% | 1,422 | 981 | -31% | 0 | 0 | — |
case-15 | pass→pass | 7,268 | 3,618 | -50% | 1 | 1 | 0% | 1,426 | 1,197 | -16% | 0 | 0 | — |
case-16 | fail→pass | 9,650 | 8,652 | -10% | 1 | 1 | 0% | 1,712 | 1,272 | -26% | 0 | 0 | — |
case-17 | pass→pass | 2,651 | 11,224 | +323% | 1 | 1 | 0% | 461 | 904 | +96% | 0 | 0 | — |
case-18 | pass→pass | 6,339 | 3,184 | -50% | 1 | 1 | 0% | 1,123 | 1,052 | -6% | 0 | 0 | — |
case-19 | fail→pass | 8,411 | 2,192 | -74% | 1 | 1 | 0% | 1,459 | 991 | -32% | 0 | 0 | — |
case-20 | pass→pass | 5,457 | 3,661 | -33% | 1 | 1 | 0% | 1,150 | 1,263 | +10% | 0 | 0 | — |
case-21 | pass→pass | 9,202 | 4,452 | -52% | 1 | 1 | 0% | 1,666 | 1,352 | -19% | 0 | 0 | — |
case-22 | pass→pass | 12,429 | 11,585 | -7% | 1 | 1 | 0% | 1,154 | 1,390 | +20% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 20 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/3/2026 | +27% |
Other measured skills in the registry, with their headline benchmark lift.