Skip to main content
skill Claude Code CX Codex Cursor GitHub Copilot

review-delta — Claude Code skill: review changes

This skill reviews changes since the last commit and maps their blast radius. Use it for focused code review without loading the whole project.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add tirth8205/code-review-graph --skill review-delta -a claude-code
Open on GitHub

What review-delta does

review-delta reads changes from git diff against HEAD~1, scores changed functions for risk, and reports test gaps and affected flows. You can pass a file or function name to narrow the review scope.

The skill starts with get_minimal_context_tool. When the graph is not ready, it calls build_or_update_graph_tool, then uses detect_changes_tool and, when needed, get_review_context_tool. For each untested high-risk function, it checks related tests with query_graph_tool.

The report includes a one-line summary, risk level, issues, impacted files and functions, plus recommendations. Whole files are loaded only when a snippet cannot answer the question. A project graph must be available or built first, so its state affects the first review step.

What it helps with

  • review only changes since the last commit
  • find high-risk functions with a wide blast radius
  • spot missing tests around changed code
  • check which tests cover an untested function
  • report impacted files, functions, and flows

Example prompts for the agent

  • Run the `review-delta` skill on changes since the last commit.
  • Use `review-delta` for `src/auth.ts` and show high-risk functions with their blast radius.
  • Run `review-delta` and verify test coverage for every untested high-risk function.

How to install review-delta

1Skills CLI (Vercel)

npx skills add tirth8205/code-review-graph --skill review-delta -a claude-code

2Manually: put the files in place

~/.claude/skills/review-delta/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/review-delta/

From the same set7

The whole set →

Similar

vuln-scan official anthropics Static vulnerability scanner for source code. Reads a directory, spawns parallel review subagents per focus area, and writes findings to JSON and markdown for triage. Claude Code CX Codex Cursor +1 1,100installs test-driven-development Hermes Agent (adapted from obra/superpowers) Skill that embeds TDD into your workflow: write the test first, watch it fail, then code. Forces you to see failing tests before implementation, reducing bugs and enabling safe refactoring. Claude Code CX Codex Cursor +1 477installs printing-press-retro mvanhorn Runs a structured retrospective after a Printing Press CLI generation run, redacts sensitive data, and drafts GitHub issues for core tool defects. Claude Code CX Codex Cursor +1 6,500installs opencode Hermes Agent Skill to run OpenCode CLI from Claude Code, Cursor, and other IDEs. Delegates refactoring, feature development, and PR review to a separate AI agent with its own session and model choice. Claude Code CX Codex Cursor +1 475installs git-workflow agno-team A skill that enforces best practices for commits, branch names, and pull requests. Includes commit message validation, branch naming templates, and PR checklist structure — useful if your team has git standards. Claude Code CX Codex Cursor +1 665installs plan-design-review garrytan Senior product designer review for plans before implementation begins. It spots missing UX decisions and updates the plan document directly. Claude Code CX Codex Cursor +1 474installs improve-codebase-architecture Matt Pocock This skill scans a codebase for architectural friction, then builds a visual HTML report with refactoring candidates. Use it before a deeper architecture review. Claude Code 1,163,035installs tdd Matt Pocock Guides the red-green TDD loop: ensures tests are written first, keep them at public interfaces, and avoid coupling to implementation details. Claude Code 1,141,062installs

Questions about review-delta

How do I install review-delta in Claude Code?

Run these steps in order: npx skills add tirth8205/code-review-graph --skill review-delta -a claude-code, then run npx skills add tirth8205/code-review-graph --skill review-delta -a claude-code again, and check ~/.claude/skills/review-delta/SKILL.md.

What does review-delta do?

It reviews changes against HEAD~1, scores their risk and blast radius, finds test gaps, and reports affected code and flows.

What happens when the project graph is not ready?

The skill calls build_or_update_graph_tool() and continues the change analysis. It uses query_graph_tool to confirm relationships such as tests for a function.

Source text

SKILL.md · MIT
---
name: review-delta
description: Review only changes since last commit using impact analysis. Token-efficient delta review with automatic blast-radius detection.
argument-hint: "[file or function name]"
---

# Review Delta

Review only the changed code and its blast radius.

## Steps

1. Call `get_minimal_context_tool(task="review changes")`. If it returns `status: not_ready`, call `build_or_update_graph_tool()` and continue.
2. Call `detect_changes_tool(detail_level="minimal")` for risk-scored changed functions, test gaps and affected flows. Changes come from `git diff` against `HEAD~1`; if the argument names a file, pass it in `changed_files`.
3. Call `get_review_context_tool(detail_level="minimal")` when you need source snippets for the changed areas and `review_guidance` (untested functions, wide blast radius, inheritance changes).
4. For each untested high-risk function, confirm with `query_graph_tool(pattern="tests_for", target="<function>")`.
5. Use `detail_level="standard"` only for a high-risk item the minimal output leaves unclear. Do not load whole files unless a snippet is not enough.

## Report

- Summary: one line
- Risk: low, medium or high, from the blast radius
- Issues: bugs, missing tests, style
- Blast radius: impacted files and functions
- Recommendations

© tirth8205 · MIT · tirth8205/code-review-graph

See also

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor