DD6 Examples — Real-World Scoring

Examples

DD6 in practice.

Each vector maps to a discovery depth recommendation, with reasoning for each dimension.

Bug

Feature

Platform

I1 D1 S1 T1 P1 B1 → 6 Skip

Fix typo in error message

I1 intent is obvious D1 no domain expertise needed S1 single stakeholder T1 trivially testable P1 strong precedent B1 boundaries clear

✓ Skip discovery ✓ Generate spec from intake

I1 D1 S1 T1 P1 B1 → 6 Skip

Add dark mode toggle to settings page

I1 clear feature request D1 standard UI pattern S1 single stakeholder T1 toggle works or it doesn't P1 well-known pattern B1 fully bounded

✓ Skip discovery ✓ Agent generates spec directly

I1 D2 S1 T1 P1 B1 → 7 Skip

Migrate from PostgreSQL 14 to 16

I1 well-defined upgrade D2 DB expertise helpful S1 engineering only T1 testable migration P1 documented upgrade path B1 clear scope

✓ Skip discovery ✓ Template-based spec

I1 D2 S1 T1 P2 B1 → 8 Shallow

Fix Safari WebSocket disconnect on idle timeout

I1 clear bug report D2 browser-specific expertise needed S1 engineering only T1 reproducible test case P2 some prior patterns B1 bounded to Safari

✓ 1–2 sessions ⚠ Expert review recommended

I2 D2 S2 T2 P2 B2 → 12 Standard

Real-time notifications for governance violations

I2 needs clarification on rules D2 governance + notification systems S2 product + engineering alignment T2 testable but rules need defining P2 similar systems exist B2 multiple delivery channels

⚠ 2–4 sessions needed ⚠ Stakeholder input required

I1 D2 S1 T1 P2 B1 → 8 Shallow

Migrate from Redis to Valkey for session caching

I1 clear migration goal D2 infrastructure expertise S1 engineering only T1 testable with existing suite P2 API-compatible but verify B1 bounded scope

✓ 1–2 sessions ⚠ Checklist + BDD scenarios

I3 D3 S1 T2 P3 B3 → 15 Deep

Intermittent data corruption in batch processing

I3 vague — root cause unknown D3 deep system internals S1 engineering-led investigation T2 hard to reproduce reliably P3 no prior occurrence B3 boundaries unknown

× 4+ sessions required × Continuous collaboration × OST + AI Bubbles

I2 D3 S2 T2 P3 B2 → 14 Deep

Add multi-tenant data isolation

I2 goal clear, approach unclear D3 security + architecture expertise S2 product + security alignment T2 testable but complex scenarios P3 no precedent in codebase B2 cross-cutting but scoped

× 4+ sessions required × Architecture exploration × Hypothesis testing

I2 D3 S3 T2 P3 B3 → 16 Deep

Decompose monolith user service into microservices

I2 direction clear, scope unclear D3 deep architecture knowledge S3 multiple teams must align T2 integration testing complex P3 first decomposition attempt B3 boundaries undefined

× 4+ sessions across teams × Multi-session iteration × OST + AI Bubbles

Template

Copy-paste scoring template.

Use this template when scoring an intake with your team.

Intake: [describe the intake]

I: _ (1-3)  Intent — how clear is what the requester wants?
D: _ (1-3)  Domain — how much specialist knowledge is needed?
S: _ (1-3)  Stakeholder — how many perspectives must converge?
T: _ (1-3)  Testability — can an agent verify the outcome?
P: _ (1-3)  Precedent — have we done something like this before?
B: _ (1-3)  Boundary — are the problem edges defined?

Score: _  (I + D + S + T + P + B)
Depth: _  (skip / shallow / standard / deep / emergency)