Claude Code Plugins

Community-maintained marketplace

Feedback

Create minimal, surgical patch plans for targeted fixes. Use when fixing specific issues from review, creating focused patches, or avoiding scope creep in fix implementations.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name patch-design
description Create minimal, surgical patch plans for targeted fixes. Use when fixing specific issues from review, creating focused patches, or avoiding scope creep in fix implementations.
allowed-tools Read, Grep

Patch Design Skill

Create minimal, surgical patch plans for targeted fixes.

When to Use

  • Fixing specific issues from review
  • Creating focused bug fixes
  • Addressing blockers without scope creep
  • Making targeted improvements

Core Principle

"This is a PATCH - keep the scope minimal. Only fix what's described in the issue and nothing more."

Design Workflow

Step 1: Understand the Specific Issue

Parse the issue clearly:

  • What is broken? (specific behavior)
  • How should it work? (expected behavior)
  • What evidence exists? (screenshots, errors)

Do NOT expand scope beyond the reported issue.

Step 2: Analyze Current State

Review the relevant code:

# Find relevant files
git diff --stat
grep -r "related_function" src/

# Read the specific code
cat src/component/file.ts

Step 3: Determine Minimum Fix

Ask: "What is the smallest change that fixes this?"

Approach Lines Risk
Surgical fix 1-10 Low
Targeted fix 10-50 Medium
Refactor 50+ High

Always prefer surgical fixes for patches.

Step 4: Create Patch Plan

Document the precise fix:

# Patch: [Concise title]

## Issue Summary
**Problem**: [What's broken]
**Solution**: [Minimal fix]

## Files to Modify
- `path/to/file.ts`: [Specific change]

## Implementation Steps

### Step 1: [Specific action]
[Exact code change]

### Step 2: [Specific action]
[Exact code change]

## Validation
- [Command to verify fix]
- [How to confirm success]

## Patch Scope
- Lines of code: ~X
- Risk level: low
- Testing: [minimal/standard]

Step 5: Define Validation

Every patch needs verification:

## Validation

1. Run specific test: `npm test path/to/test`
2. Manual verification: [steps]
3. Visual check: [if applicable]

Patch Plan Template

# Patch: [Brief description]

## Metadata
- Issue: [Source of this patch request]
- Severity: [blocker/tech_debt]

## Issue Summary
**Original Spec**: [Link if available]
**Problem**: [Brief description of what's wrong]
**Solution**: [Brief description of the fix]

## Files to Modify
[Only files that need changes - be specific]

- `src/component/Button.tsx`: Fix disabled state

## Implementation Steps

IMPORTANT: Execute every step in order.

### Step 1: [Specific change]

<!-- markdownlint-disable MD033 -->

```jsx
// Before
<button onClick={handleClick}>

// After
<button onClick={handleClick} disabled={isLoading}>

Step 2: [Next change if needed]

[Details]

Validation

Execute to confirm patch is complete:

  • npm test src/component/Button.test.tsx
  • Manual: Click button, verify disabled during load

Patch Scope

  • Lines changed: ~5
  • Risk level: low
  • Testing: Run component tests

Anti-Patterns to Avoid

Scope Creep

# BAD: Expanding scope
While fixing the button, I noticed the form could use
refactoring, so I'll also update the validation logic,
improve the error messages, and add loading states to
all other buttons too.

Vague Changes

# BAD: Vague plan
## Implementation
- Fix the button issue
- Make it work correctly
- Test to confirm

No Validation

# BAD: No verification
## Implementation
1. Change the code
2. Done

Best Practices

  1. One issue, one patch: Don't bundle fixes
  2. Minimal scope: Only change what's needed
  3. Specific steps: Exact code changes
  4. Clear validation: How to verify success
  5. Risk assessment: Lines changed, testing needed

Integration with Review

Patches typically come from review issues:

/review → Identifies blocker
  ↓
/patch → Creates surgical fix plan
  ↓
/implement → Executes patch
  ↓
/review → Verifies fix (re-review)

Memory References

  • @issue-severity-classification.md - What triggers patches
  • @review-vs-test.md - Review identifies patch needs
  • @one-agent-one-purpose.md - Patches are focused by nature

Version History

  • v1.0.0 (2025-12-26): Initial release

Last Updated

Date: 2025-12-26 Model: claude-opus-4-5-20251101