Pre-Flight Code Check
Structured Claude skill that gives Claude a repeatable pre-flight discipline check before coding.
Before & after
Same prompt, same agent — the only difference is the skill.
Before shipping, you should definitely review your code carefully. Some things to think about: Is the logic correct? Are there any edge cases you missed? Do you have enough error handling? What about performance — are there any inefficient loops or queries? Make sure your variable names are clear and your functions do one thing each. Have you tested the main paths? Consider security too. I'd recommend going through each file and asking yourself whether it makes sense. It's easy to miss things, so taking time here saves a lot of pain later. Would you like to walk through the code together?
PRE-FLIGHT CODE CHECK → LOGIC: Edge cases identified → boundary conditions tested for null/empty/negative inputs; loops properly bounded; conditionals cover all branches → NAMING: Variable and function names unmistakable; no abbreviations or single-letter vars outside loops; intent clear from first read → ERROR HANDLING: Try/catch blocks present where needed; failures logged with context; graceful fallbacks instead of silent failures → PERFORMANCE: No nested loops on large datasets; database queries indexed; string concatenation replaced with proper builders; image sizes optimized → SECURITY: No hardcoded credentials; input validated and sanitized; SQL injection prevented via prepared statements; CORS headers set correctly → TESTING: Happy path tested; at least one failure case per function; boundary values checked; async operations awaited → DEPENDENCIES: All imports used; no circular dependencies; version pins reasonable; no known vulnerabilities → DEPLOYMENT: Environment variables all set; database migrations run; logs at trace level; rollback plan documented
About this skill
name: karpathy-check description: Use when Structured Claude skill that gives Claude a repeatable workflow for the original author check.
Pre-Flight Code Check
One of 337+ skills in the original author's multi-agent claude-skills mega-collection (~19k GitHub stars). Packages the the original author Check workflow with its own instructions and validation so outputs stay consistent.
What you get
- Public GitHub repo (alirezarezvani/claude-skills)
- the the original author-check skill folder with SKILL.md. Part of a 337-skill / 30-agent / 70-command install.
Customize your output
- Fork the repo and adapt the skill's instructions and references to your workflow.
Example output
Activates automatically when your request matches the original author Check; chains with the other skills, agents, and commands in the collection.
Best for
Creators, builders, and teams using Claude Code.
SKILL.md preview
---
name: karpathy-check
description: Use this skill before starting a coding task, as a pre-flight discipline check that confirms the problem is actually understood before any code gets written.
version: 1.0.0
category: AI Agents / Development
author: AgentVolt
license: proprietary
tags:
- ai-agents
- development
---
# Pre-Flight Code Check
Runs a short discipline check before coding starts: confirm the problem is understood, the assumptions are verified, and "done" is defined, so effort doesn't get spent solving the wrong thing.
## When to use
… (sign up to view the full skill)More ai & agents skills
View all AI & Agents skills →Skill Status Report
Scan every skill in a project and report which pass validation, which are stale, and which lack required metadata.
Skills Scaffolder
Scaffold a new skills directory — folder structure, metadata, a baseline validation pass — with the right conventions from day one.
Research Bundle
Run a multi-step research task as one pipeline: gather sources, extract findings, validate consistency, emit one structured bundle.
Workspace Admin
Administer a Cowork or Claude workspace — settings, access, configuration — with every change scoped and verified.