moai-workflow-thinking

Performs structured step-by-step reasoning via the Sequential Thinking MCP tool.

2|Updated May 27, 2026
One-click install
npx skills add https://github.com/yekinya/moai-novel --skill moai-workflow-thinking-yekinya
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: moai-workflow-thinking
Source: https://github.com/yekinya/moai-novel/tree/main/moai-novel/.claude/skills/moai-workflow-thinking
Command: npx skills add https://github.com/yekinya/moai-novel --skill moai-workflow-thinking-yekinya

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex architecture decisions and multi-step problems often get resolved with undocumented, ad-hoc reasoning that cannot be reviewed or revised. This Skill enforces a recorded, step-by-step thinking chain through the Sequential Thinking MCP tool when the --deepthink flag is used. ## Core Features & Use Cases - Structured Thought Chains: Invokes mcp__sequential-thinking__sequentialthinking with numbered thoughts, total estimates, and revision tracking. - Branching and Revision: Supports revising earlier thoughts and branching alternative approaches for decisions with multiple viable paths. - Mode Disambiguation: Clearly separates --deepthink (MCP-based) from ultrathink (Claude native extended reasoning) and Opus 4.7 Adaptive Thinking. - Use Case: When evaluating a breaking change or choosing between technologies affecting 3+ files, trigger --deepthink to produce a documented reasoning chain with revisions and a verifiable conclusion. ## Quick Start Ask the AI to analyze an architecture decision using the --deepthink flag so it produces a step-by-step sequential thinking chain with a documented conclusion.

Frequently Asked Questions about moai-workflow-thinking

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I use sequential thinking for architecture decisions?

Pass the --deepthink flag so the Sequential Thinking MCP tool is invoked with numbered thoughts, a totalThoughts estimate, and revision steps. The chain ends when a final thought sets nextThoughtNeeded to false with a conclusion.

What is the difference between --deepthink and ultrathink?

--deepthink invokes the Sequential Thinking MCP tool for recorded step-by-step analysis, while ultrathink uses Claude's native extended reasoning without any MCP tool. They are independent modes and can be activated together.

Does sequential thinking work with GLM models?

No. The Sequential Thinking MCP tool generates a server_tool_use content type that GLM does not support. Use ultrathink or Adaptive Thinking instead, since they rely on native reasoning without special content types.

When should I use branching in sequential thinking?

Use branching when two or more viable alternatives exist for the same problem. Set branchFromThought and branchId to explore each path explicitly so the final conclusion can compare documented alternatives.

What are the limitations of sequential thinking on Opus 4.7?

On Opus 4.7, Adaptive Thinking is the built-in reasoning mode and overrides fixed-budget instructions. Do not hardcode a reasoning budget; let the model adapt depth automatically based on task complexity.