antipattern-detector

Anti-Pattern Detector

Identifies recurring failure patterns in plans and decisions. Anti-patterns look reasonable on the surface but lead to predictable problems. Detecting them early saves months of pain.

Anti-Pattern Categories

1. Architecture Anti-Patterns

Pattern	Description	Symptoms
Big Ball of Mud	No clear structure, everything coupled	Can't change X without breaking Y
Golden Hammer	Using one approach for everything	"We'll use [tool] for that too"
Premature Microservices	Splitting before understanding boundaries	3 people managing 20 services
Distributed Monolith	Microservices with tight coupling	Must deploy all services together
Resume-Driven Development	Choices for career, not product	"Let's use Rust for the admin panel"

2. Timeline Anti-Patterns

Pattern	Description	Symptoms
Timeline Fantasy	Optimistic estimates ignoring reality	"6 weeks if everything goes well"
Scope Creep Blindness	Not accounting for inevitable growth	Same deadline, 2x scope
Parallel Path Delusion	Assuming unlimited parallelization	"Add more people to go faster"
MVP Maximalism	MVP that's actually V3	47 features in "minimum" product
Demo-Driven Development	Building for demos, not production	"It works on my machine"

3. Team Anti-Patterns

Pattern	Description	Symptoms
Hero Culture	Reliance on key individuals	"Only Sarah can fix that"
Knowledge Silos	Critical info in one person's head	Bus factor of 1
Conway's Law Violation	Structure doesn't match team	Boundaries don't align
Understaffed Ambition	Big plans with tiny teams	2 people building "the platform"
Absent Ownership	No clear owner for components	Issues fall through cracks

4. Process Anti-Patterns

Pattern	Description	Symptoms
Cargo Cult Agile	Ceremonies without principles	Standups but no shipping
Analysis Paralysis	Over-planning, under-executing	Month 3 of "finalizing requirements"
Infinite Refactoring	Never shipping, always improving	"One more cleanup before release"
Documentation Theater	Docs that no one reads	200-page spec, outdated day 1
Meeting Madness	More meetings than doing	"Let's schedule a meeting"

5. Technology Anti-Patterns

Pattern	Description	Symptoms
Shiny Object Syndrome	Chasing latest tech without reason	"Rewrite in [new hotness]"
Not Invented Here	Building what should be bought	Custom auth, custom logging, custom everything
Vendor Lock-in Denial	Ignoring exit costs	"We can always migrate later"
Premature Optimization	Optimizing before measuring	Caching layer with 10 users
Framework Overload	Too many frameworks/libraries	47 dependencies for a button

---

Detection Process

Step 1: Scan for Signals

Look for phrases that indicate anti-patterns:

Timeline signals: "If everything goes well...", "We can do it faster if...", "Just need to hire...", "Should only take..."

Architecture signals: "We'll figure out boundaries later...", "Everything talks to everything...", "It's only temporary...", "We can always refactor..."

Team signals: "Only [person] knows...", "We'll hire for that...", "[Person] will handle all of...", "The team can absorb..."

Process signals: "We don't need docs...", "We'll add tests later...", "Let's discuss in the meeting...", "Requirements are evolving..."

Step 2: Verify Pattern Match

For each suspected anti-pattern:

1. Identify: Which specific pattern?
2. Evidence: What in the plan matches?
3. Severity: How bad? (Critical / High / Medium / Low)
4. Context: Could this be a reasonable exception?

Step 3: Check for Combinations

Certain patterns appear together and compound risk:

Startup Death Spiral: Timeline Fantasy + Understaffed Ambition + Hero Culture

• Result: Burnout, missed deadlines, key person leaves

Enterprise Trap: Analysis Paralysis + Documentation Theater + Meeting Madness

• Result: Nothing ships, team frustrated, competition wins

Tech Debt Avalanche: "Refactor later" + No ownership + Premature optimization

• Result: System unmaintainable, rewrite required

Microservices Mistake: Premature Microservices + Distributed Monolith + Understaffed

• Result: Complexity explosion, slower than monolith

---

Severity Framework

Level	Meaning	Action
Critical	Will cause failure if not addressed	Stop and fix before proceeding
High	Will cause significant problems	Address in planning phase
Medium	Will cause friction and delays	Include in risk mitigation
Low	Worth noting but manageable	Track and address opportunistically

---

When to Use This Skill

Use antipattern-detector for recognizing known failure patterns (architecture, timeline, team, process, technology). Use assumption-challenger for surfacing hidden assumptions treated as facts. Both run together in /validate.

Pre-Delivery Checklist

Before presenting an analysis, verify:

• Every detected pattern includes a direct evidence quote from the plan
• Every pattern has a severity rating (Critical/High/Medium/Low)
• Pattern combinations are checked and reported
• "Patterns NOT Detected" section is included for confidence
• Fixes are specific and actionable, not generic advice

---

Output Format

# Anti-Pattern Analysis: [Plan Name]

## Summary
- **Patterns Detected**: [Count]
- **Critical**: [Count] | **High**: [Count] | **Medium**: [Count] | **Low**: [Count]
- **Overall Risk Level**: Critical / High / Medium / Low
- **Pattern Combinations Found**: [List any compound patterns]

## Critical Issues (Must Address)

### [Pattern Name]
**Category**: [Category] | **Severity**: Critical
**Evidence**: [What triggered detection]
**Risk**: [What will go wrong]
**Fix**: [How to address]

---

## High-Priority Issues (Should Address)
[Same format]

## Medium-Priority Issues (Consider Addressing)
[Same format]

## Patterns NOT Detected
[List patterns checked but not found — builds confidence]

## Recommendations
### Before Proceeding
1. [Critical action]
### During Execution
1. [Mitigation]
### Warning Signs to Monitor
- [Metric or signal to watch]