tech-debt

name	tech-debt
model	claude-sonnet-4-0
description	Technical debt analysis and remediation. Scans codebase for code smells, duplication, complexity, testing gaps, and architecture issues. Generates prioritized reports with ROI estimates. Use /tech-debt to run full analysis or /tech-debt [area] for targeted scan.
argument-hint	[area] e.g. challenges, goals, all

Technical Debt Analysis and Remediation

Analyze the HeadsUp codebase to identify, quantify, and prioritize technical debt. Generate actionable remediation plans with clear ROI.

Target: $ARGUMENTS

---

Workflow

Phase 1: Scan & Inventory

Conduct a thorough scan based on the target area. If no area specified, scan the full codebase.

Target Mapping:

Argument	Scan Scope
all	Full codebase
challenges	lib/heads_up/challenges*, lib/heads_up_web/live/challenge_live/*
goals	lib/heads_up/goals*, lib/heads_up_web/live/goal_live/*
users	lib/heads_up/accounts*, lib/heads_up_web/live/users_live/*
social	Social features: likes, comments, subscriptions, feed
admin	lib/heads_up_web/live/admin_live/*
tests	test/**/*_test.exs
schemas	lib/heads_up/**/*.ex (Ecto schemas only)
liveviews	lib/heads_up_web/live/**/*.ex

1.1 Code Debt Scan

Search for these patterns:

Duplicated Code

• Grep for identical function bodies across files
• Look for repeated query patterns in contexts
• Find duplicated template/component markup
• Identify copy-pasted event handlers in LiveViews

Complex Code

• Functions longer than 50 lines
• Deeply nested case/cond/if blocks (>3 levels)
• LiveView render/1 functions over 200 lines
• Context modules over 500 lines
• Functions with more than 5 parameters

Poor Structure

• Context functions that bypass the context pattern (direct Repo calls in LiveViews)
• Missing ownership/authorization checks
• Circular module dependencies
• Business logic in templates/components

1.2 Architecture Debt Scan

• Missing context boundaries (logic in wrong context)
• Fat LiveViews (business logic in event handlers instead of contexts)
• Missing PubSub for real-time features
• Inconsistent error handling patterns
• Missing or inconsistent API boundaries

1.3 Testing Debt Scan

• Files with no corresponding test file
• Context functions without test coverage
• LiveView events without test coverage
• Missing edge case tests (empty states, boundaries, errors)
• Slow or flaky tests

1.4 Schema & Migration Debt

• Missing indexes on foreign keys
• Missing unique constraints
• Inconsistent timestamp types (naive_datetime vs utc_datetime)
• Missing on_delete specifications
• Schema fields without validation in changeset

1.5 Documentation Debt

• Public context functions without @doc
• Missing @moduledoc on modules
• Undocumented complex business rules
• Missing typespec (@spec) on public functions

---

Phase 2: Impact Assessment

For each debt item found, assess:

Severity Levels:

• Critical: Security risk, data integrity, production crashes
• High: Bugs likely, significant dev time waste, performance degradation
• Medium: Developer friction, maintenance burden, inconsistency
• Low: Code style, minor inefficiency, cosmetic

Impact Scoring:

Impact = Frequency_of_change x Severity x Blast_radius

Where:
  Frequency: How often this code is touched (1-5)
  Severity: How bad the debt is (1-5)
  Blast_radius: How much code is affected (1-5)

---

Phase 3: Generate Report

Create the report at project/tech-debt/TD-[DATE]-[area].md

# Technical Debt Report: [Area]
**Date:** YYYY-MM-DD
**Scope:** [what was scanned]
**Overall Score:** X/100 (lower = more debt)

## Executive Summary
- Total debt items: X
- Critical: X | High: X | Medium: X | Low: X
- Estimated remediation effort: X hours
- Top 3 priorities: [list]

## Debt Inventory

### Critical Items
| # | Category | File | Description | Impact | Effort |
|---|----------|------|-------------|--------|--------|
| 1 | [type]   | path | description | score  | hours  |

### High Priority Items
[same table format]

### Medium Priority Items
[same table format]

### Low Priority Items
[same table format]

## Metrics

### Code Quality
- Average function length: X lines
- Functions >50 lines: X
- Modules >500 lines: X
- Duplicated patterns: X locations

### Test Coverage
- Context functions tested: X%
- LiveView events tested: X%
- Files without tests: X

### Schema Health
- Missing indexes: X
- Missing validations: X
- Inconsistent patterns: X

## Quick Wins (< 4 hours each)
1. [item] - Effort: Xh, Savings: Xh/month
2. [item] - Effort: Xh, Savings: Xh/month

## Remediation Roadmap

### Week 1-2: Quick Wins
- [ ] [task]
- [ ] [task]

### Month 1: High Priority
- [ ] [task]
- [ ] [task]

### Month 2-3: Medium Priority
- [ ] [task]
- [ ] [task]

## Prevention Recommendations
1. [recommendation]
2. [recommendation]

---

Phase 4: Create Tasks

For each Quick Win and High Priority item, create a task file:

Location: project/tech-debt/tasks/TDT-XXX-[slug].md

# TDT-XXX: [Task Title]

## Source
- Report: TD-[DATE]-[area].md
- Severity: [Critical/High/Medium/Low]
- Impact Score: X/125

## Description
[What needs to change and why]

## Files Affected
- `path/to/file.ex`

## Steps
1. [step]
2. [step]
3. [step]

## Effort Estimate
- Hours: X
- Complexity: [Low/Medium/High]

## Acceptance Criteria
- [ ] [criterion]
- [ ] [criterion]
- [ ] Tests pass: `mix test`

## Status
- [ ] Planned
- [ ] In Progress
- [ ] Done
- [ ] Verified

---

Elixir/Phoenix Specific Checks

Context Pattern Violations

# BAD: Repo call in LiveView
def handle_event("save", params, socket) do
  Repo.insert(changeset)  # Should be in context
end

# GOOD: Context function
def handle_event("save", params, socket) do
  Goals.create_goal(params)
end

LiveView Anti-patterns

# BAD: Business logic in template
<%= if length(user.goals) >= 2 do %>  # Magic number, logic in template

# GOOD: Helper or context function
<%= if Goals.at_limit?(@current_user) do %>

Schema Checks

# Check for: missing validates, missing indexes, missing on_delete
schema "goals" do
  belongs_to :user, User  # Should have on_delete
  field :status, Ecto.Enum  # Should validate in changeset
end

Query Performance

# BAD: N+1 query
goals |> Enum.map(fn g -> Repo.preload(g, :user) end)

# GOOD: Batch preload
goals |> Repo.preload(:user)

---

Output

After generating the report, display a summary:

Tech Debt Analysis Complete
============================
Scope: [area]
Report: project/tech-debt/TD-[DATE]-[area].md
Tasks: X created in project/tech-debt/tasks/

Summary:
  Critical: X items
  High:     X items
  Medium:   X items
  Low:      X items
  Total:    X items

Top 3 Quick Wins:
  1. [description] (Xh effort, Xh/month savings)
  2. [description] (Xh effort, Xh/month savings)
  3. [description] (Xh effort, Xh/month savings)

Estimated total remediation: X hours