Skip to content

DirectoryKits

Fix a Bug

Find the root cause before anyone touches the code.

BuildFree to startUpdated

Three parts, one job

Each part works on its own. In this kit each one hands the next a file.

Use

Cursor

Its agent reads the project, edits the code and runs the tests.

Takes
An error or a failing test you can make happen again
Hands on
error.log
How
Ask Cursor's agent to rerun the failing command and save the whole output, stack trace included, to error.log in your project.

Give your agent

Systematic Debugging

Makes the agent find the root cause before it proposes a fix.

Takes
error.log
Hands on
debug-notes.md
How
Paste the trace, code and steps from debug-notes.md into [Error message and stack trace], [Relevant code] and [What you were doing].

Run

Debugging for Errors and Stack Traces

The root cause, a quick way to confirm it and the fix, from your stack trace.

Takes
debug-notes.md
Gives you
The proven root cause, a quick check and the fix to make in Cursor

Use any part on its own, or all three in this order.

Set it up

In the order you use them.

1

Use Cursor

Ask Cursor's agent to rerun the failing command and save the whole output, stack trace included, to error.log in your project.

Visit CursorFreecursor.com

Or use GitHub Copilot or Claude Code; keep the skill and the prompt.

cursor.com
Cursor homepage
2

Give your agent Systematic Debugging

Paste the trace, code and steps from debug-notes.md into [Error message and stack trace], [Relevant code] and [What you were doing].

$/plugin install superpowers@claude-plugins-official
By @obraMIT297k stars
obra/superpowers › systematic-debugging/SKILL.md

---

name: systematic-debugging

description: Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

---

# Systematic Debugging

## Overview

**Core principle:** ALWAYS find root cause before attempting fixes.

Symptom fixes are failure.

**Violating the letter of this process is violating the spirit of

3

Run Debugging for Errors and Stack Traces

Fill in the 3 blanks, then run it.

Error message and stack traceRelevant codeWhat you were doing
Fill it in
recaply.co/prompts/coding
<error_and_stack_trace> {{Error message and stack trace}} </error_and_stack_trace> <code> {{Relevant code}} </code> <what_happened> {{What you were doing}} </what_happened> </context> <task> Using the error, the code and my notes above, find the root cause and give me a fix. A good answer: - names the line that throws and, separately, the line or input where the problem starts, and shows how one leads to the other - accounts for every clue in my notes, including the cases where the error does not happen - fixes the cause rather than only silencing the error - tells me how to confirm the cause in a minute or two before I change an

Questions about this kit

Is this kit free?

Yes, to start. Cursor has a free plan, and the skill and the prompt are free.

Is there a tool other than Cursor for this job?

Yes. GitHub Copilot and Claude Code do the same job well. Keep the skill and the prompt.

How does my agent get Systematic Debugging?

Run /plugin install superpowers@claude-plugins-official in Claude Code. Your agent reads the skill when the task fits.

Where is the Debugging for Errors and Stack Traces prompt?

On its page: fill in the blanks, then copy it or open it in ChatGPT, Claude or Grok.

Who picks the kits?

We do. A sponsor can take the labelled row under a kit, but never one of its three picks.

More kits in Build

Pages, features and fixes, with your agent.

Build a Landing Page: quick look

Build a Landing Page

From one sentence to a live page, without a template look.

  1. Use Base44
  2. Give your agent Frontend Design
  3. Run Claude Code to Add a Feature
Ship a Feature: quick look

Ship a Feature

Think it through first, then let your agent write tested code.

  1. Use Claude Code
  2. Give your agent Brainstorming
  3. Run Write a Function From a Spec
Review Code Before You Merge: quick look

Review Code Before You Merge

Your branch reviewed, the findings fixed and a description reviewers can follow.

  1. Use CodeRabbit
  2. Give your agent Code Review and Quality
  3. Run Pull Request Description
Secure Your App: quick look

Secure Your App

Scan the code, fix what it finds, then check what scanners miss before launch.

  1. Use Snyk
  2. Give your agent Security and Hardening
  3. Run Security Check for Vibe-Coded Apps