# MCP Sequential Thinking

> A detailed tool for dynamic and reflective problem-solving through thoughts that can adapt and evolve

- Skill: `jr2804/mcp-sequential-thinking` (Agent Skill)
- Install (CLI): `npx skillmds@latest add jr2804/mcp-sequential-thinking`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jr2804/mcp-sequential-thinking/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: jr2804 (https://skillmd.com/u/jr2804)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/jr2804/mcp-sequential-thinking

---


# Sequential Thinking

**If MCP server `sequential-thinking` is available**, use this tool to analyze complex problems through a flexible thinking process.

## When to Use

Use sequential-thinking when:

- Breaking down complex problems into steps
- Planning and design with room for revision
- Analysis that might need course correction
- Problems where full scope might not be clear initially
- Problems that require a multi-step solution
- Tasks that need to maintain context over multiple steps
- Situations where irrelevant information needs to be filtered out

## Key Features

- **Flexible thinking process**: You can adjust total_thoughts up or down as you progress
- **Revision support**: You can question or revise previous thoughts
- **Continuation**: You can add more thoughts even after reaching what seemed like the end
- **Uncertainty expression**: You can express uncertainty and explore alternative approaches
- **Non-linear**: Not every thought needs to build linearly - you can branch or backtrack
- **Hypothesis-driven**: Generates a solution hypothesis and verifies it

## Parameters

```python
thought: str  # Your current thinking step
nextThoughtNeeded: (
    bool  # True if you need more thinking, even if at what seemed like the end
)
thoughtNumber: int  # Current number in sequence (can go beyond initial total if needed)
totalThoughts: int  # Current estimate of thoughts needed (can be adjusted up/down)
isRevision: bool  # A boolean indicating if this thought revises previous thinking
revisesThought: (
    int  # If is_revision is true, which thought number is being reconsidered
)
branchFromThought: int  # If branching, which thought number is the branching point
branchId: str  # Identifier for the current branch (if any)
needsMoreThoughts: bool  # If reaching end but realizing more thoughts needed
```

## Workflow

01. **Start with initial estimate**: Begin with an estimate of needed thoughts, but be ready to adjust
02. **Question and revise**: Feel free to question or revise previous thoughts
03. **Don't hesitate to add thoughts**: Add more thoughts if needed, even at "end"
04. **Express uncertainty**: Express uncertainty when present
05. **Mark revisions clearly**: Mark thoughts that revise previous thinking or branch into new paths
06. **Ignore irrelevant info**: Ignore information that is irrelevant to current step
07. **Generate hypothesis**: Generate a solution hypothesis when appropriate
08. **Verify hypothesis**: Verify hypothesis based on the Chain of Thought steps
09. **Repeat until satisfied**: Repeat the process until satisfied with a solution
10. **Provide final answer**: Provide a single, ideally correct answer as final output
11. **Set nextThoughtNeeded to false**: Only when truly done and a satisfactory answer is reached

## Example Thought Progression

```text
Thought 1: Analyzing the problem requirements...
Thought 2: Considering possible approaches...
Thought 3: Revising approach from Thought 2 due to X constraint...
Thought 4: Generating solution hypothesis...
Thought 5: Verifying hypothesis...
```

## Best Practices

- **Be systematic**: Follow a structured thinking process
- **Be adaptive**: Allow yourself to revise course based on new information
- **Be explicit**: Clearly mark when you're revising or branching
- **Be thorough**: Don't skip verification steps
- **Be concise**: Each thought should be focused and specific

## References

- [Sequential Thinking GitHub](https://github.com/modelcontextprotocol/servers/tree/main/src/sequentialthinking)
- [Model Context Protocol](https://github.com/modelcontextprotocol/servers)