Analysis Skill

Conducts in-depth analysis on a specified topic and produces a comprehensive markdown report saved to docs/analysis/.

Arguments

The skill takes a single argument: the analysis prompt describing what to analyze.

Examples:

• •/analyze the test coverage gaps in the iOS codebase
• •/analyze authentication security patterns used in the backend
• •/analyze performance bottlenecks in the question generation service
• •/analyze code duplication across the project

Execution Steps

Step 1: Parse the Analysis Request

Extract the topic from the user's prompt. The topic should describe what needs to be analyzed.

Step 2: Generate Output Filename

Create a descriptive filename based on the topic:

• •Convert to lowercase
• •Replace spaces with hyphens
• •Remove special characters
• •Add timestamp prefix for uniqueness
• •Example: 2024-01-15-ios-test-coverage-gaps.md

# Generate timestamp
TIMESTAMP=$(date +%Y-%m-%d)
# The filename will be: docs/analysis/${TIMESTAMP}-<slugified-topic>.md

Step 3: Conduct the Analysis

Based on the topic, gather relevant information by:

1. •Searching the codebase using Glob and Grep tools to find relevant files and patterns
2. •Reading key files to understand implementation details
3. •Using specialized agents via the Task tool when needed:
   • •Use Explore agent for codebase exploration
   • •Use domain-specific agents for specialized analysis (e.g., statistical-analysis-scientist for statistical topics)
4. •Web research if the topic requires external context or best practices comparison

Step 4: Structure the Analysis

The analysis report should follow this structure:

# Analysis: <Topic Title>

**Date:** <YYYY-MM-DD>
**Scope:** <Brief description of what was analyzed>

## Executive Summary

<2-3 paragraph high-level summary of findings>

## Methodology

<How the analysis was conducted>
- Tools and techniques used
- Files and directories examined
- External resources consulted (if any)

## Findings

### <Finding Category 1>

<Detailed findings with evidence>

#### Evidence
- File references with line numbers
- Code snippets where relevant
- Metrics or statistics

### <Finding Category 2>

<Continue with additional findings...>

## Recommendations

| Priority | Recommendation | Effort | Impact |
|----------|---------------|--------|--------|
| High | ... | ... | ... |
| Medium | ... | ... | ... |
| Low | ... | ... | ... |

### Detailed Recommendations

#### 1. <Recommendation Title>

**Problem:** <What issue this addresses>
**Solution:** <Proposed approach>
**Files Affected:** <List of files>

## Appendix

### Files Analyzed
<List of all files that were examined>

### Related Resources
<Links to documentation, best practices, etc.>

Step 5: Write the Analysis File

Write the complete analysis to docs/analysis/<filename>.md:

# Ensure directory exists
mkdir -p docs/analysis

Then use the Write tool to create the file at docs/analysis/${TIMESTAMP}-<topic-slug>.md.

Step 6: Summarize and Report

After writing the file, provide the user with:

1. •The path to the generated analysis file
2. •A brief summary of key findings
3. •The top 3 recommendations

Quality Standards

• •Thoroughness: Examine all relevant areas of the codebase
• •Evidence-based: Support all findings with specific file references and code examples
• •Actionable: Recommendations should be concrete and implementable
• •Organized: Use clear headings and consistent formatting
• •Balanced: Present both strengths and areas for improvement

Example Output Summary

After completing the analysis, report to the user:

Analysis complete. Report saved to: docs/analysis/2024-01-15-ios-test-coverage-gaps.md

Key Findings:
1. Test coverage is 45% overall, with ViewModels at 62% and Services at 38%
2. No tests exist for 12 critical authentication flows
3. UI snapshot tests are missing for all custom components

Top Recommendations:
1. [High] Add unit tests for AuthenticationService (estimated 15 test cases)
2. [High] Implement snapshot tests for reusable UI components
3. [Medium] Increase ViewModel test coverage to 80%

Important Guidelines

1. •Be comprehensive: Don't rush the analysis - examine all relevant code paths
2. •Cite sources: Always include file paths and line numbers for findings
3. •Prioritize findings: Not all issues are equally important
4. •Stay focused: Keep the analysis relevant to the requested topic
5. •Use agents appropriately: Delegate specialized analysis to domain experts via Task tool

Begin by parsing the analysis request and planning the investigation approach.