@gotza02/ultimatethinking

The Ultimate MCP Server for Sequential Thinking

Advanced Sequential Thinking MCP Server for dynamic and reflective problem-solving. Features session management, goal tracking, branching, and security enhancements.

Features

• Sequential Thinking: Break down complex problems into manageable steps
• Analysis Modes: Tailor reasoning for debugging, decisions, comparisons, planning, and research
• Session Management: Create and manage multiple thinking sessions
• Goal Tracking: Set goals with success criteria and milestones
• Progress Tracking: Automatic progress calculation
• Branching: Explore alternative paths of reasoning
• Revision Support: Revise and refine previous thoughts
• Analytical Structure: Capture assumptions, evidence, alternatives, criteria, risks, and open questions
• Quality Feedback: Surface quality score, missing elements, unresolved questions, and analysis flags
• Thought Categories: analysis, hypothesis, conclusion, revision, verification, insight, question, action
• Confidence Levels: Track confidence (0-100) for each thought
• Tags & References: Organize and link thoughts together

Security

• Prototype Pollution Protection: Blocked dangerous branch IDs
• Memory Limits: Configurable limits for history, branches, and sessions
• Input Validation: Comprehensive Zod schema validation

Installation

Option 1: Install globally

npm install -g @gotza02/ultimatethinking

Option 2: Use with npx (no installation required)

npx @gotza02/ultimatethinking

Usage with MCP Clients

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

Using npx (Recommended)

{
  "mcpServers": {
    "ultimatethinking": {
      "command": "npx",
      "args": ["-y", "@gotza02/ultimatethinking"]
    }
  }
}

Using global install

{
  "mcpServers": {
    "ultimatethinking": {
      "command": "ultimatethinking"
    }
  }
}

Using local path

{
  "mcpServers": {
    "ultimatethinking": {
      "command": "node",
      "args": ["/path/to/ultimatethinking/dist/index.js"]
    }
  }
}

Kimi-CLI

Edit ~/.kimi/mcp.json:

{
  "mcpServers": {
    "ultimatethinking": {
      "command": "npx",
      "args": ["-y", "@gotza02/ultimatethinking"]
    }
  }
}

Gemini-CLI

Edit ~/.gemini/settings.json:

{
  "mcpServers": {
    "ultimatethinking": {
      "command": "npx",
      "args": ["-y", "@gotza02/ultimatethinking"]
    }
  }
}

Available Tools

1. sequential_thinking (Primary)

Main thinking tool for problem-solving.

| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | thought | string | Yes | Current thinking step | | nextThoughtNeeded | boolean | Yes | Whether more thinking needed | | thoughtNumber | integer | Yes | Current thought number (1, 2, 3...) | | totalThoughts | integer | Yes | Estimated total thoughts | | category | enum | No | analysis, hypothesis, conclusion, revision, verification, insight, question, action | | analysisMode | enum | No | general, debugging, decision, comparison, planning, research | | confidence | integer | No | Confidence level (0-100) | | isRevision | boolean | No | Whether this revises previous thinking | | revisesThought | integer | No | Which thought is being reconsidered | | branchFromThought | integer | No | Branching point thought number | | branchId | string | No | Branch identifier | | references | string[] | No | IDs of referenced thoughts in the current session | | assumptions | string[] | No | Explicit assumptions behind the thought | | evidence | string[] | No | Evidence supporting the thought | | alternatives | string[] | No | Alternatives considered | | criteria | string[] | No | Decision criteria or evaluation dimensions | | risks | string[] | No | Risks, downsides, or failure modes | | openQuestions | string[] | No | Outstanding questions that remain unresolved | | resolvedQuestions | string[] | No | Questions resolved by this thought in the same scope | | tags | string[] | No | Tags for categorization |

Hard validation rules:

• references must point to existing thought IDs in the current session before they count as support
• resolvedQuestions only count when they match questions that are already open in the same scope
• Branch thoughts are analyzed in branch scope; non-branch thoughts are analyzed in mainline scope
• hypothesis thoughts must include assumptions, evidence, or references
• verification thoughts must include evidence, references, or resolved questions
• final decision, comparison, and planning thoughts must include alternatives and criteria
• final general, debugging, and research thoughts must include evidence or references
• final thoughts must provide openQuestions explicitly, using [] when nothing remains

2. create_thinking_session

Create a new thinking session with optional goal.

3. set_session_goal

Set or update goal for current session.

4. update_goal_progress

Update a specific criterion or milestone and recompute goal progress.

5. get_session_info

Get current session statistics, progress, and analytical summary. Accepts optional branchId to return branch-scoped analysis and branch metadata.

6. get_thought_history

Get all thoughts from current session. Accepts optional branchId to return only thoughts from that branch.

7. clear_session

Clear all thoughts and reset goal progress/completion state while keeping the session active.

Configuration

| Setting | Default | Description | |---------|---------|-------------| | maxHistorySize | 1000 | Max thoughts per session | | maxBranches | 100 | Max branches per session | | maxSessions | 10 | Max concurrent sessions |

System Instruction for AI

See SYSTEM_INSTRUCTION.md for detailed configuration.

Quick System Prompt

You have access to @gotza02/ultimatethinking MCP.
For complex questions requiring analysis, planning, or reasoning:
USE sequential_thinking FIRST.
Do NOT ask permission - use it automatically.
Use analysisMode plus assumptions/evidence/alternatives when relevant.
Use update_goal_progress when updating tracked criteria or milestones.
Only respond AFTER completing your thinking process.

Development

# Clone
git clone https://github.com/gotza02/ultimatethinking.git
cd ultimatethinking

# Install
npm install

# Build
npm run build

# Run locally
npm start
# or
node dist/index.js

Architecture

src/
├── index.ts           # MCP server entry point
├── session-manager.ts # Session & thought management
├── types.ts           # TypeScript definitions
├── schemas.ts         # Zod validation schemas
└── utils.ts           # Utilities & LRU cache

Publishing

# Login to npm
npm login

# Publish
npm publish --access public

License

MIT © gotza