iterative-depth

name: iterative-depth description: Multi-angle deep exploration through cognitive lenses, each pass surfaces requirements and insights invisible from other angles. USE WHEN iterative depth, multi-angle, deep exploration, multiple passes, lenses, thorough analysis, edge cases, hidden requirements.

Iterative Depth: Multi-Angle Deep Exploration

Structured multi-angle exploration of the same problem through cognitive lenses. Each pass surfaces requirements, edge cases, and insights invisible from other angles. Grounded in 20 established techniques across cognitive science, AI/ML, requirements engineering, and design thinking.

Customization

Check for user customizations at: ~/.augment/USER/SKILLCUSTOMIZATIONS/IterativeDepth/PREFERENCES.md

If this file exists, load and apply preferences found there. They override defaults below.

Core Concept

Instead of analyzing a problem once, run 2-8 structured passes through the same problem, each from a systematically different lens. The combination yields understanding that no single-pass analysis can produce.

Why it works:

1. Perspective Blindness Compensation -- Any single viewpoint has blind spots. Rotating through viewpoints covers gaps.
2. Progressive Pre-Understanding -- Each pass updates understanding, making subsequent passes more perceptive.
3. Convergence Signal -- When multiple lenses surface the same insight, it is high-priority.

The 8 Lenses

Ordered from most concrete to most abstract, and from most commonly useful to most specialized.

Lens 1: LITERAL (Surface Requirements)

Question: "What did they explicitly say? What are the concrete, stated requirements?" Extract every stated requirement, constraint, preference. No interpretation -- only what was said.

Lens 2: STAKEHOLDER (Who Else Cares?)

Question: "Who are all the people, systems, and entities affected? What does each need?" Identify every stakeholder beyond the requester: end users, maintainers, admins, downstream systems, future developers.

Lens 3: FAILURE (What Goes Wrong?)

Question: "What could fail? What would an adversary exploit? What are the edge cases?" Assume the solution exists. Break it. Error states, race conditions, security holes, performance under load.

Lens 4: TEMPORAL (Past, Present, Future)

Question: "How does this change over time? What happens in 6 months?" Why does this problem exist now? What changes in the future that would break this solution? Migration paths, backwards compatibility.

Lens 5: EXPERIENTIAL (How Should It Feel?)

Question: "When this works perfectly, how does the user FEEL?" Beyond functional correctness: speed, elegance, confidence, trust. The difference between "works" and "works beautifully."

Lens 6: CONSTRAINT INVERSION (What If?)

Question: "What if we removed all constraints? What if we added extreme ones?" Remove assumed constraints -- what would we build? Add extreme constraints -- what is truly essential? Both reveal hidden assumptions.

Lens 7: ANALOGICAL (What Patterns Apply?)

Question: "What similar problems have been solved before? What patterns from other domains apply?" This problem is not unique. What emerged in other codebases, industries, fields?

Lens 8: META (Is This the Right Question?)

Question: "Are we solving the right problem? Is the framing itself correct?" Step outside the problem. Is the request a symptom of a deeper issue? Would a different question yield a better outcome?

Depth Selection

When invoked with a specific count (e.g., "do 3 passes"), select lenses in order from Lens 1 through Lens N.

Domain-Specific Lens Priority

Execution Process

Step 1: Determine Depth

Based on complexity, user request, or stakes. Default to Standard (4 lenses) if unclear.

Step 2: Run Passes Sequentially

For each lens, produce:

ITERATIVE DEPTH -- Pass [i]/[N]: [LENS NAME]

Lens Question: "[The lens's core question]"

Findings:
- [Finding 1 -- potential criterion or insight]
- [Finding 2 -- potential criterion or insight]
- [Finding 3 -- refinement of something found in earlier pass]

New Insights:
+ [New insight this lens uniquely reveals]
~ [Refined insight -- was X, now Y based on this angle]

Step 3: Synthesize

After all passes:

1. Deduplicate -- Remove semantically identical findings across lenses
2. Merge refinements -- If multiple lenses refined the same finding, take the most specific version
3. Prioritize -- Order by how many lenses surfaced the insight (consensus = high priority)
4. Identify the key insight -- The most surprising finding that single-pass analysis would have missed

Output Format

## Iterative Depth Analysis: [Topic]

### Coverage
- Lenses used: [list]
- New insights discovered: [count]
- Existing insights refined: [count]

### Findings by Lens
[Summary of each lens pass with key findings]

### Consolidated Insights (Priority Order)
1. [Highest priority -- surfaced by multiple lenses]
2. [Second priority]
3. [Third priority]
...

### Key Insight
[The most surprising finding across all lenses -- the thing single-pass analysis would have missed]

Scientific Foundation

This technique draws from validated methods across disciplines:

Anti-Patterns

• Repetitive Passes: Each lens must be structurally different, not just a re-run with slight variation
• Skipping Synthesis: Running all lenses but not consolidating defeats the purpose
• Surface-Only: Using lenses as checklists rather than genuinely exploring from each angle
• Over-Depth: Using 8 lenses on a simple task that needs 2 -- match depth to stakes
• Ignoring Convergence: When multiple lenses find the same thing, it is a strong signal -- do not dismiss it