Fix - Bug Fix Workflow

Fix a specific bug or problem in the codebase. Supports two modes: immediate fix or plan-first approach.

Workflow

Step 0: Check for Existing Fix Plan

BEFORE anything else, check if .ai-factory/FIX_PLAN.md exists.

If the file EXISTS:

• •Read .ai-factory/FIX_PLAN.md
• •Inform the user: "Found existing fix plan. Executing fix based on the plan."
• •Skip Steps 0.1 through 1 — go directly to Step 2: Investigate the Codebase, using the plan as your guide
• •Follow each step of the plan sequentially
• •After the fix is fully applied and verified, delete .ai-factory/FIX_PLAN.md:
  bash

  rm .ai-factory/FIX_PLAN.md
  

• •Continue to Step 4 (Verify), Step 5 (Test suggestion), Step 6 (Patch)

If the file DOES NOT exist AND $ARGUMENTS is empty:

• •Tell the user: "No fix plan found and no problem description provided. Please either provide a bug description (/ai-factory.fix <description>) or create a fix plan first."
• •STOP.

If the file DOES NOT exist AND $ARGUMENTS is provided:

• •Continue to Step 0.1 below.

Step 0.1: Load Project Context & Past Experience

Read .ai-factory/DESCRIPTION.md if it exists to understand:

• •Tech stack (language, framework, database)
• •Project architecture
• •Coding conventions

Read all patches from .ai-factory/patches/ if the directory exists:

• •Use Glob to find all *.md files in .ai-factory/patches/
• •Read each patch file to learn from past fixes
• •Pay attention to recurring patterns, root causes, and solutions
• •If the current problem resembles a past patch — apply the same approach or avoid the same mistakes
• •This is your accumulated experience. Use it.

Step 1: Understand the Problem & Choose Mode

From $ARGUMENTS, identify:

• •Error message or unexpected behavior
• •Where it occurs (file, function, endpoint)
• •Steps to reproduce (if provided)

If unclear, ask:

To fix this effectively, I need more context:

1. What is the expected behavior?
2. What actually happens?
3. Can you share the error message/stack trace?
4. When did this start happening?

After understanding the problem, ask the user to choose a mode using AskUserQuestion:

Question: "How would you like to proceed with the fix?"

Options:

1. •Fix now — Investigate and apply the fix immediately
2. •Plan first — Create a fix plan for review, then fix later

If user chooses "Plan first":

• •Proceed to Step 1.1: Create Fix Plan

If user chooses "Fix now":

• •Skip Step 1.1, proceed directly to Step 2: Investigate the Codebase

Step 1.1: Create Fix Plan

Investigate the codebase enough to understand the problem and create a plan.

1. •Search for relevant files using Glob/Grep
2. •Read the code around the issue
3. •Trace the data flow
4. •Identify the root cause (or most likely candidates)

Then create .ai-factory/FIX_PLAN.md with this structure:

# Fix Plan: [Brief title]

**Problem:** [What's broken — from user's description]
**Created:** YYYY-MM-DD HH:mm

## Analysis

What was found during investigation:
- Root cause (or suspected root cause)
- Affected files and functions
- Impact scope

## Fix Steps

Step-by-step plan for implementing the fix:

1. [ ] Step one — what to change and why
2. [ ] Step two — ...
3. [ ] Step three — ...

## Files to Modify

- `path/to/file.ts` — what changes are needed
- `path/to/another.ts` — what changes are needed

## Risks & Considerations

- Potential side effects
- Things to verify after the fix
- Edge cases to watch for

## Test Coverage

- What tests should be added
- What edge cases to cover

After creating the plan, output:

## Fix Plan Created ✅

Plan saved to `.ai-factory/FIX_PLAN.md`.

Review the plan and when you're ready to execute, run:

/ai-factory.fix

STOP here. Do NOT apply the fix.

Step 2: Investigate the Codebase

Search for the problem:

• •Find relevant files using Glob/Grep
• •Read the code around the issue
• •Trace the data flow
• •Check for similar patterns elsewhere

Look for:

• •The root cause (not just symptoms)
• •Related code that might be affected
• •Existing error handling

Step 3: Implement the Fix

Apply the fix with logging:

// ✅ REQUIRED: Add logging around the fix
console.log('[FIX] Processing user input', { userId, input });

try {
  // The actual fix
  const result = fixedLogic(input);
  console.log('[FIX] Success', { userId, result });
  return result;
} catch (error) {
  console.error('[FIX] Error in fixedLogic', {
    userId,
    input,
    error: error.message,
    stack: error.stack
  });
  throw error;
}

Logging is MANDATORY because:

• •User needs to verify the fix works
• •If it doesn't work, logs help debug further
• •Feedback loop: user provides logs → we iterate

Step 4: Verify the Fix

• •Check the code compiles/runs
• •Verify the logic is correct
• •Ensure no regressions introduced

Step 5: Suggest Test Coverage

ALWAYS suggest covering this case with a test:

## Fix Applied ✅

The issue was: [brief explanation]
Fixed by: [what was changed]

### Logging Added
The fix includes logging with prefix `[FIX]`.
Please test and share any logs if issues persist.

### Recommended: Add a Test

This bug should be covered by a test to prevent regression:

\`\`\`typescript
describe('functionName', () => {
  it('should handle [the edge case that caused the bug]', () => {
    // Arrange
    const input = /* the problematic input */;

    // Act
    const result = functionName(input);

    // Assert
    expect(result).toBe(/* expected */);
  });
});
\`\`\`

Would you like me to create this test?
- [ ] Yes, create the test
- [ ] No, skip for now

Logging Requirements

All fixes MUST include logging:

1. •Log prefix: Use [FIX] or [FIX:<issue-id>] for easy filtering
2. •Log inputs: What data was being processed
3. •Log success: Confirm the fix worked
4. •Log errors: Full context if something fails
5. •Configurable: Use LOG_LEVEL if available

// Pattern for fixes
const LOG_FIX = process.env.LOG_LEVEL === 'debug' || process.env.DEBUG_FIX;

function fixedFunction(input) {
  if (LOG_FIX) console.log('[FIX] Input:', input);

  // ... fix logic ...

  if (LOG_FIX) console.log('[FIX] Output:', result);
  return result;
}

Examples

Example 1: Null Reference Error

User: /ai-factory.fix TypeError: Cannot read property 'name' of undefined in UserProfile

Actions:

1. •Search for UserProfile component/function
2. •Find where .name is accessed
3. •Add null check with logging
4. •Suggest test for null user case

Example 2: API Returns Wrong Data

User: /ai-factory.fix /api/orders returns empty array for authenticated users

Actions:

1. •Find orders API endpoint
2. •Trace the query logic
3. •Find the bug (e.g., wrong filter)
4. •Fix with logging
5. •Suggest integration test

Example 3: Form Validation Not Working

User: /ai-factory.fix email validation accepts invalid emails

Actions:

1. •Find email validation logic
2. •Check regex or validation library usage
3. •Fix the validation
4. •Add logging for validation failures
5. •Suggest unit test with edge cases

Important Rules

1. •Check FIX_PLAN.md first - Always check for existing plan before anything else
2. •Plan mode = plan only - When user chooses "Plan first", create the plan and STOP. Do NOT fix.
3. •Execute mode = follow the plan - When FIX_PLAN.md exists, follow it step by step, then delete it
4. •NO reports - Don't create summary documents (patches are learning artifacts, not reports)
5. •ALWAYS log - Every fix must have logging for feedback
6. •ALWAYS suggest tests - Help prevent regressions
7. •Root cause - Fix the actual problem, not symptoms
8. •Minimal changes - Don't refactor unrelated code
9. •One fix at a time - Don't scope creep
10. •Clean up - Delete FIX_PLAN.md after successful fix execution

After Fixing

## Fix Applied ✅

**Issue:** [what was broken]
**Cause:** [why it was broken]
**Fix:** [what was changed]

**Files modified:**
- path/to/file.ts (line X)

**Logging added:** Yes, prefix `[FIX]`
**Test suggested:** Yes

Please test the fix and share logs if any issues.

To add the suggested test:
- [ ] Yes, create test
- [ ] No, skip

Step 6: Create Self-Improvement Patch

ALWAYS create a patch after every fix. This builds a knowledge base for future fixes.

Create the patch:

1. •

   Create directory if it doesn't exist:

   bash

   mkdir -p .ai-factory/patches
   

2. •

   Create a patch file with the current timestamp as filename. Format: YYYY-MM-DD-HH.mm.md (e.g., 2026-02-07-14.30.md)

3. •

   Use this template:

# [Brief title describing the fix]

**Date:** YYYY-MM-DD HH:mm
**Files:** list of modified files
**Severity:** low | medium | high | critical

## Problem

What was broken. How it manifested (error message, wrong behavior).
Be specific — include the actual error or symptom.

## Root Cause

WHY the problem occurred. This is the most valuable part.
Not "what was wrong" but "why it was wrong":
- Logic error? Why was the logic incorrect?
- Missing check? Why was it missing?
- Wrong assumption? What was assumed?
- Race condition? What sequence caused it?

## Solution

How the fix was implemented. Key code changes and reasoning.
Include the approach, not just "changed line X".

## Prevention

How to prevent this class of problems in the future:
- What pattern/practice should be followed?
- What should be checked during code review?
- What test would catch this?

## Tags

Space-separated tags for categorization, e.g.:
`#null-check` `#async` `#validation` `#typescript` `#api` `#database`

Example patch:

# Null reference in UserProfile when user has no avatar

**Date:** 2026-02-07 14:30
**Files:** src/components/UserProfile.tsx
**Severity:** medium

## Problem

TypeError: Cannot read property 'url' of undefined when rendering
UserProfile for users without an uploaded avatar.

## Root Cause

The `user.avatar` field is optional in the database schema but the
component accessed `user.avatar.url` without a null check. This was
introduced in commit abc123 when avatar display was added — the
developer tested only with users that had avatars.

## Solution

Added optional chaining: `user.avatar?.url` with a fallback to a
default avatar URL. Also added a null check in the Avatar sub-component.

## Prevention

- Always check if database fields marked as `nullable` / `optional`
  are handled with null checks in the UI layer
- Add test cases for "empty state" — user with minimal data
- Consider a lint rule for accessing nested optional properties

## Tags

`#null-check` `#react` `#optional-field` `#typescript`

This is NOT optional. Every fix generates a patch. The patch is your learning.

Context Cleanup

Context is heavy after investigation, fix, and patch generation. All results are saved — suggest freeing space:

AskUserQuestion: Free up context before continuing?

Options:
1. /clear — Full reset (recommended)
2. /compact — Compress history
3. Continue as is

DO NOT:

• •❌ Apply a fix when user chose "Plan first" — only create FIX_PLAN.md and stop
• •❌ Skip the FIX_PLAN.md check at the start
• •❌ Leave FIX_PLAN.md after successful fix execution — always delete it
• •❌ Generate reports or summaries (patches are NOT reports — they are learning artifacts)
• •❌ Refactor unrelated code
• •❌ Add features while fixing
• •❌ Skip logging
• •❌ Skip test suggestion
• •❌ Skip patch creation