Claude-code-customes smart-think

Advanced multi-mode thinking system with Sequential Thinking MCP and Serena integration for complex problem solving

install

source · Clone the upstream repo

git clone https://github.com/sc30gsw/claude-code-customes

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/sc30gsw/claude-code-customes "$T" && mkdir -p ~/.claude/skills && cp -r "$T/.claude/skills/smart-think" ~/.claude/skills/sc30gsw-claude-code-customes-smart-think && rm -rf "$T"

manifest: .claude/skills/smart-think/SKILL.md

source content

Smart Think: Advanced Multi-Mode Problem Solving

Intelligent problem-solving with multiple thinking modes, budget control, and MCP integration for complex analysis and decision-making.

Usage

/smart-think "<problem_description>" [options]

Options

Option	Short	Description	Default
`--mode`	`-m`	Thinking mode	`think`
`--budget`	`-b`	Token budget	Auto
`--interactive`	`-i`	Interactive refinement	false
`--output`	`-o`	Save to file	none
`--serena`	`-s`	Use Serena integration	false
`--research`	`-r`	Include research phase	false
`--citations`	`-c`	Include citations	false
`--structured`		Structured output	false
`--confidence`		Show confidence levels	false
`--focus`	`-f`	Focus area	none

Thinking Modes

Mode	Budget Range	Thoughts	Confidence	Best For
`think`	2,000-8,000	3-6	70-85%	Quick decisions
`think-hard`	8,000-15,000	6-10	75-90%	Complex analysis
`think-harder`	15,000-25,000	10-15	80-95%	Architecture decisions
`ultrathink`	25,000-50,000	15-25	85-98%	Research, complex systems

Tool Priorities

ALWAYS prioritize mcp__sequential-thinking__ as primary engine, with mcp__serena__ for codebase context:

Primary Thinking Engine (Sequential Thinking MCP)

Use

mcp__sequential-thinking__sequentialthinking

as primary thinking tool

Generate and test multiple solution hypotheses
Build reasoning chains with confidence tracking

Codebase Intelligence (Serena MCP - When --serena flag)

Use
```
mcp__serena__get_symbols_overview
```
for technical understanding
Use
```
mcp__serena__search_for_pattern
```
for implementation insights

Use

mcp__serena__read_memory

mcp__serena__write_memory

for decision continuity

Example Usage

# Default thinking mode
/smart-think "Should we use Redux or Zustand?"

# Deep analysis with codebase context
/smart-think "Database migration strategy" -m think-harder --serena

# Research-focused with citations
/smart-think "Technology selection" -m ultrathink --research --citations

# Quick decision with budget control
/smart-think "CSS framework choice" -b 5000 --focus=frontend

Mode Selection Guidelines

Think: Quick decisions, simple problems, time constraints
Think-Hard: Important decisions, moderate complexity
Think-Harder: Critical decisions, high complexity
UltraThink: Mission-critical, research-level analysis

Output Formats

Default Format

Problem Analysis
Solution Exploration
Recommendation
Next Steps

Research Format (with --research)

Executive Summary
Literature Review
Technical Analysis
Recommendations
References

Technical Format (with --serena)

Technical Context
Code Analysis
Architecture Implications
Implementation Strategy