high-level-code-review

Reviews code changes for semantic correctness, completeness, consistency, and design quality.

3.3k|1.1k|Updated Nov 24, 2014
One-click install
npx skills add https://github.com/stellar/stellar-core --skill high-level-code-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: high-level-code-review
Source: https://github.com/stellar/stellar-core/tree/main/.claude/skills/high-level-code-review
Command: npx skills add https://github.com/stellar/stellar-core --skill high-level-code-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Low-level review catches mechanical mistakes, but it cannot tell whether a change actually accomplishes its intended goal or fits the codebase's design. This Skill performs a structured high-level review of a git diff, evaluating correctness, completeness, performance, consistency, safety, error handling, minimality, and documentation, then produces a prioritized worklist of issues.

Core Features & Use Cases

  • Goal-driven review: Requires the stated goal of the change and a git diff command, then evaluates whether the change actually achieves that goal.
  • Eight review criteria: Covers correctness, completeness, performance, consistency, safety, error handling, minimality, and documentation, with codebase-specific checks such as FeeBumpTransactionFrame support and deterministic randomness.
  • Structured issue report: Outputs each issue with file path, line numbers, category, severity (Critical/Major/Minor/Suggestion), description, and an actionable recommendation.
  • Use Case: Before merging a pull request that modifies transaction processing in stellar-core, run this review to verify edge cases are handled, patterns match existing code, and no unintended side effects or missing tests remain.

Quick Start

Review the change for adding fee bump support to transaction submission using the diff from git diff master...HEAD and report any correctness or completeness issues.

Frequently Asked Questions about high-level-code-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a pull request for correctness beyond style issues?

High-level code review evaluates whether a change accomplishes its stated goal by checking correctness, completeness, safety, and consistency against the surrounding codebase. Provide the change goal and a git diff command, and the review produces a severity-ranked issue list with recommendations.

What inputs does a high-level code review need?

The review requires the goal of the change (from an issue, PR description, or commit message) and a git command to obtain the diff, such as git diff master...HEAD. Optional specific concerns can focus attention on particular areas.

What is the difference between high-level and low-level code review?

Low-level review catches mechanical mistakes like typos and syntax errors, while high-level review evaluates whether the change is correct in intent and approach. It checks algorithms, edge cases, error paths, design consistency, and whether the change is minimal and complete.

Can this review be split across multiple subagents?

Yes, the review can be divided into pieces run as subagents, for example one per review criterion or one per file. Each subagent should receive the change goal, the git diff command, and a focused scope so it stays on track.

What output format does the code review produce?

The review produces a structured report listing each issue with file path, line numbers, category, severity (Critical, Major, Minor, or Suggestion), a description, and a recommendation. If no issues exist, it confirms the change appears correct with supporting observations.