Help us improve
Share bugs, ideas, or general feedback.
From claude-leverage
Reads a git diff and produces a plain-English narration of changes for PR descriptions, code reviews, or self-orientation. Supports three audience modes: --for pr, --for review, and --for self.
npx claudepluginhub filip-podstavec/claude-leverage --plugin claude-leverageHow this skill is triggered — by the user, by Claude, or both
Slash command
/claude-leverage:explain-diff [ref-range, default HEAD] [--for pr|review|self][ref-range, default HEAD] [--for pr|review|self]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Reads a git diff and produces a tight English narration of what
Generates structured PR descriptions from git commit history by grouping changes by domain, filtering noise like lints, and inspecting diffs only when needed. Use for consistent, token-efficient PR summaries.
Visualizes git diffs, branches, commits, PRs, and ranges as interactive HTML reports with architecture diagrams, KPI dashboards, code review cards, and side-by-side comparisons.
Reviews recent git diffs and commits with brutal honesty before PRs, spotting 2am logic flaws, copy-paste artifacts, debug leftovers, hacks, and poor naming.
Share bugs, ideas, or general feedback.
Reads a git diff and produces a tight English narration of what
changed, why a reviewer should care, and where the risk concentrates.
Lives next to the vanilla commit workflow (mechanical Conventional
Commits message) and /security-review (structured security audit) —
covers the "narrative" niche neither does.
Three audience modes:
--for pr (default) — output is a PR-description-shaped block:
## Summary, ## Why, ## How to verify. Copy-pasteable into the
PR body.--for review — emphasizes what's load-bearing and what's mechanical;
flags non-obvious decisions that the reviewer might miss without
the author's voice ("the new IF EXISTS clause on line 47 is
load-bearing — without it the migration is non-idempotent").--for self — terse 3–5 bullet self-narration; useful before
switching context or coming back to a branch later.Resolve diff range.
git diff HEAD (working tree + staged vs last commit).$ARGUMENTS starts with ^ or contains .. or ..., treat
as a ref range and use git diff <range>.--for pr and the current branch has commits beyond main,
use git diff main...HEAD instead (covers the whole branch).Get supporting context (parallel-OK):
git status --short for file-state breakdown.git log --oneline -10 for repo style of commit messages.git rev-parse --abbrev-ref HEAD for branch name.Read the diff in full. If the diff is enormous (>50k tokens
estimated), STOP and report: "diff is too large — narrow with
--paths <pattern> or <file>...". Don't try to chunk and
summarize partial diffs; the result would be incoherent.
Walk the diff and identify:
*_test.*, *.test.*,
tests/, etc.).auth*, crypto*,
*.env*, routes/, etc.) — surface as a "consider /security-review"
reminder, not a substitute for it.Emit narration in the requested shape.
--for pr output template## Summary
- <one sentence per key change, ordered by importance>
## Why
<one paragraph: the motivation. If the diff doesn't make the motivation
obvious, say so explicitly — "motivation not inferable from the diff;
add to PR body".>
## How to verify
- [ ] <reviewer step 1>
- [ ] <reviewer step 2>
- <commands to run if applicable>
--for review output template## What changed (load-bearing)
- `file:line` — <description of load-bearing change>
- ...
## What changed (mechanical, low-risk)
- <bulk list, less detail>
## What to look at first
<top 1–2 things the reviewer should NOT skip>
## What's untested
- <files modified without corresponding test changes, IF the language
conventionally tests, IF the file is non-trivial>
--for self output template- <bullet 1, ≤80 chars>
- <bullet 2>
- <bullet 3>
[max 5]
routes/auth/... — consider
/security-review". Don't try to be that skill.services/old/*
to services/new/*").--for pr output into the PR body.--for review to make sure the diff is honestly readable.--for self to
re-orient./security-review do the actual audit.main automatically in --for self / --for review
modes. Default is HEAD (working tree vs last commit). User
picks the range explicitly when they want branch-wide.