القائمة
Use when analyzing code cognitive load, onboarding difficulty, or user invokes /cognitive-audit for complexity analysis and refactoring recommendations
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use when researching state-of-the-art algorithms for a problem, finding academic papers, comparing algorithm approaches, needing paper citations, or user invokes /algorithm-research
Use when researching algorithms, data structures, optimization, performance, or need modern alternatives to classic algorithms
Use when creating benchmarks, measuring performance, comparing implementations, or user invokes /benchmark. Provides step-by-step workflow for benchmark creation, data generation, and results interpretation.
Use when analyzing cognitive load, code complexity, onboarding difficulty, readability concerns, maintainability issues, or applying Ousterhout principles for deep modules, reducing complexity, and strategic programming approaches
Use when implementing concurrent operations, choosing between concurrency models, designing async workflows, or working with parallel data processing
Use when validating input data, implementing parse-don't-validate patterns, designing validation pipelines, or handling external data safely
| name | cognitive-audit |
| description | Use when analyzing code cognitive load, onboarding difficulty, or user invokes /cognitive-audit for complexity analysis and refactoring recommendations |
Workflow for analyzing code cognitive load, onboarding difficulty, and providing strategic refactoring recommendations.
Invoke with /cognitive-audit when:
Skip this skill when:
# Analyze entire codebase
/cognitive-audit
# Analyze specific module
/cognitive-audit lib/my_app/order_processor.ex
# Analyze directory
/cognitive-audit lib/my_app/payments/
# Analyze multiple modules
/cognitive-audit lib/my_app/{orders,payments,inventory}
This skill typically uses:
Load these skills when needed during analysis:
.claude/project-learnings.md (if exists).claude/cognitive-audit-report.md (previous audit, if exists)Use the Task tool to spawn a cognitive-scientist agent (Opus model) with this prompt:
Analyze cognitive complexity for: [target files/directories]
Apply Ousterhout's principles (load cognitive-complexity skill for reference):
1. Deep vs shallow modules - Calculate depth ratio (power / complexity)
2. Information leakage - Temporal coupling, pass-through, exposed internals
3. Complexity direction - Push-up vs pull-down analysis
4. Strategic vs tactical code - Identify accumulating complexity
5. Errors that could be defined out - Type-level prevention opportunities
Measure cognitive load:
- Working memory load (parameters, lifespans, concerns, nesting)
- Semantic ambiguity (generic names, inconsistent naming)
- Temporal coupling (hidden dependencies)
- Pass-through complexity (arguments through layers)
Assess onboarding:
- Context requirements (modules to understand first)
- Domain knowledge needed
- Historical decisions documented (or missing)
- State machine complexity
- Overall difficulty score (1-10)
Generate recommendations:
- Prioritized refactoring list with ROI
- Specific code changes suggested
- Estimated impact on onboarding time
Create or update these files:
.claude/cognitive-audit-report.md - Full audit findings:
# Cognitive Complexity Audit Report
Generated: [timestamp]
Scope: [files analyzed]
## Executive Summary
- **Overall complexity**: [score]/10
- **Onboarding difficulty**: [score]/10
- **Estimated comprehension time**: [hours] for new developer
## Primary Issues
### 1. [Issue Category] (Severity)
- **Module**: [name]
- **Problem**: [description]
- **Impact**: [who/what affected]
- **Recommendation**: [specific fix]
[... additional issues ...]
## Ousterhout Analysis
### Deep vs Shallow Modules
[Table of modules with depth ratios]
### Information Leakage
[List of leakage points]
### Strategic vs Tactical Code
[Assessment of code investment patterns]
## Refactoring Recommendations
### High Priority (This Sprint)
[Prioritized list with ROI]
### Medium Priority (Next Sprint)
[Secondary improvements]
## Metrics
| Metric | Current | Target | Status |
|--------|---------|--------|--------|
| Shallow modules | X | 0 | [status] |
| High working memory functions | X | <5 | [status] |
| Temporal coupling instances | X | 0 | [status] |
| Onboarding difficulty | X/10 | 4/10 | [status] |
## Next Steps
1. [Actionable item]
2. [Actionable item]
.claude/onboarding-guide-[module].md (for complex areas):
# Onboarding Guide: [Module Name]
## Prerequisites
- Understanding of [concept 1]
- Familiarity with [concept 2]
## Key Concepts
[Core concepts to understand]
## Code Flow
[How requests/data flow through this module]
## Common Gotchas
[Things that trip up new developers]
## Related Modules
[What to read next]
Update .claude/project-learnings.md with cognitive patterns section.
Present summary to user:
# Cognitive Complexity Analysis Complete
## Executive Summary
**Scope**: [files analyzed]
**Overall complexity**: [score]/10
**Onboarding difficulty**: [score]/10
## Top Issues
1. **[Issue 1]** - [impact]
2. **[Issue 2]** - [impact]
3. **[Issue 3]** - [impact]
## Recommended Actions
### High Priority
[Top 3 refactoring recommendations with ROI]
## Reports Generated
- `.claude/cognitive-audit-report.md` - Full analysis
- `.claude/onboarding-guide-[module].md` - [if created]
- `.claude/project-learnings.md` - Updated with patterns
## Next Steps
1. Review full report
2. Prioritize refactoring with team
3. Schedule strategic work
4. Re-run audit after refactoring
**Ready to proceed with any refactoring?**
Depth = Implementation Power / Interface Complexity
> 2.0 = Deep module (good!)
1.0-2.0 = Balanced (acceptable)
< 1.0 = Shallow module (refactor needed)
Score = Parameters + Lifespan + Concerns + Nesting
0-5 = Low load (easy to understand)
6-10 = Moderate load (requires focus)
11-15 = High load (difficult)
16+ = Very high load (refactor!)
1-3 = Easy (new dev productive in days)
4-6 = Moderate (new dev productive in 1-2 weeks)
7-8 = Hard (new dev productive in 3-4 weeks)
9-10 = Very hard (new dev productive in 1-2 months)
Recommended triggers:
Warning signs requiring audit:
Audit succeeds when: