sc-analyze

Analysis & Troubleshooting Skill

Multi-domain code analysis with issue diagnosis and resolution capabilities.

Quick Start

# Quality analysis
/sc:analyze [target] --focus quality|security|performance|architecture

# Troubleshooting mode
/sc:analyze [issue] --troubleshoot --focus bug|build|performance|deployment

# With auto-fix
/sc:analyze "TypeScript errors" --troubleshoot --focus build --fix

Behavioral Flow

1. Discover - Categorize source files, detect languages
2. Scan - Apply domain-specific analysis techniques
3. Evaluate - Generate prioritized findings with severity
4. Recommend - Create actionable recommendations
5. Report - Present comprehensive analysis with metrics

Flags

Flag	Type	Default	Description
--focus	string	quality	quality, security, performance, architecture, bug, build, deployment
--troubleshoot	bool	false	Enable issue diagnosis mode
--trace	bool	false	Detailed trace analysis for debugging
--fix	bool	false	Auto-apply safe fixes
--depth	string	standard	quick, standard, deep
--format	string	text	text, json, report

Analysis Domains

Quality Analysis

• Code smells and maintainability issues
• Pattern violations and anti-patterns
• Technical debt assessment

Security Analysis

• Vulnerability scanning
• Compliance validation
• Authentication/authorization review

Performance Analysis

• Bottleneck identification
• Resource utilization patterns
• Optimization opportunities

Architecture Analysis

• Component coupling assessment
• Dependency analysis
• Design pattern evaluation

Troubleshooting Mode

When --troubleshoot is enabled:

Focus	Behavior
bug	Error analysis, stack traces, code inspection
build	Build logs, dependencies, config validation
performance	Metrics analysis, bottleneck identification
deployment	Environment analysis, service validation

Examples

Security Deep Dive

/sc:analyze src/auth --focus security --depth deep

Build Failure Fix

/sc:analyze "compilation errors" --troubleshoot --focus build --fix

Performance Diagnosis

/sc:analyze "slow API response" --troubleshoot --focus performance --trace

MCP Integration

PAL MCP (Always Use)

Tool	When to Use	Purpose
mcp__pal__thinkdeep	Complex issues	Multi-stage investigation with hypothesis testing
mcp__pal__debug	Bug troubleshooting	Systematic root cause analysis
mcp__pal__codereview	Quality analysis	Comprehensive code quality, security, performance review
mcp__pal__consensus	Critical findings	Multi-model validation of security/architecture issues
mcp__pal__challenge	Uncertain findings	Force critical thinking on ambiguous issues
mcp__pal__apilookup	Dependency issues	Get current API docs for version conflicts

PAL Usage Patterns

# Deep investigation (--depth deep)
mcp__pal__thinkdeep(
    step="Investigating performance bottleneck in API layer",
    hypothesis="Database queries lack proper indexing",
    confidence="medium",
    relevant_files=["/src/api/users.py"]
)

# Security analysis (--focus security)
mcp__pal__codereview(
    review_type="security",
    findings="Authentication, authorization, injection vectors",
    issues_found=[{"severity": "high", "description": "SQL injection risk"}]
)

# Critical finding validation
mcp__pal__consensus(
    models=[
        {"model": "gpt-5.2", "stance": "for"},
        {"model": "gemini-3-pro", "stance": "against"}
    ],
    step="Evaluate: Is this a critical security vulnerability?"
)

Rube MCP (When Needed)

Tool	When to Use	Purpose
mcp__rube__RUBE_SEARCH_TOOLS	External analysis	Find security scanners, linters
mcp__rube__RUBE_MULTI_EXECUTE_TOOL	Issue tracking	Create tickets for findings
mcp__rube__RUBE_REMOTE_WORKBENCH	Bulk analysis	Process large codebases

Rube Usage Patterns

# Find and create Jira tickets for findings
mcp__rube__RUBE_SEARCH_TOOLS(queries=[
    {"use_case": "create jira issue", "known_fields": "project:SECURITY"}
])

# Notify team of critical findings
mcp__rube__RUBE_MULTI_EXECUTE_TOOL(tools=[
    {"tool_slug": "SLACK_SEND_MESSAGE", "arguments": {"channel": "#security", "text": "Critical finding..."}}
])

Flags (Extended)

Flag	Type	Default	Description
--pal-deep	bool	false	Use PAL thinkdeep for multi-stage analysis
--pal-review	bool	false	Use PAL codereview for comprehensive review
--consensus	bool	false	Use PAL consensus for critical findings
--notify	string	-	Notify via Rube (slack, jira, email)
--create-tickets	bool	false	Create tickets for findings via Rube

Tool Coordination

• Glob - File discovery and structure analysis
• Grep - Pattern analysis and code search
• Read - Source inspection and config analysis
• Bash - External tool execution
• Write - Report generation
• PAL MCP - Multi-model analysis, debugging, code review
• Rube MCP - External notifications, ticket creation