mcp-sequentialthinking-tools

A lightweight MCP server for step-by-step reasoning: a scratchpad with history, branching and tool-plan validation

  • 588 ★ on the repository
  • Last commit
  • MIT
MCP server

Extended access

Why this level

  • Runs as a separate process and stores step history
  • Handles text treated as untrusted with best-effort filtering
All reasons and checks

spences10/mcp-sequentialthinking-tools

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add mcp/mcp-sequentialthinking-tools

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Runs as a separate process and stores step history
  • Handles text treated as untrusted with best-effort filtering

This is third-party code. Review the repository files before installing.

What it does

The server gives the model a place to record its reasoning step by step. Each step is a thought with a number, an estimate of the total, and a flag for whether another step is needed. History can be kept per session, a thought can be marked as a revision, and branches can start from it. If you pass the available tools and the recommended ones for a step, the server checks that the recommended names exist and otherwise returns an error. It does not discover your tools or choose them for the model. Thought and description text is treated as untrusted: suspicious injections are scrubbed before storage.

Who it is for. For people who want to see and check an agent's reasoning on hard multi-step tasks.

Good fit when

  • You need to break a messy task into explicit steps
  • You need to revise or branch a plan as work goes
  • You need to check that a plan references only existing tools

Not a fit when

  • The task is trivial and an extra reasoning layer only gets in the way
  • You expect an external reasoning engine, this is only a scratchpad with validation

Example request

Break this task into steps and keep a reasoning history so I can review it

MCP

Transport
stdio
Authentication
not required

Environment variables

MAX_HISTORY_SIZE
Per-session reasoning history size, default 1000

Limitations

The server does not expose your other tools or choose them for the model, it only stores and validates steps. The suspicious-text filtering is best-effort, not a guarantee, so do not put secrets in thoughts. History size is capped by MAX_HISTORY_SIZE.

How to disable. Remove the mcp-sequentialthinking-tools entry from the MCP client config.

FAQ

Does it pick tools for the model?

No. The server only stores steps and, if you pass the lists, checks that the recommended tool names are among the available ones.

Is it a reasoning engine?

No, it is a scratchpad with history, branching and validation. It does not think for the model, it records and validates its steps.

Discussion

No questions or reviews yet.

A local MCP server that keeps a staged journal of an agent's reasoning with persistent sessions, decisions and evidence

MCP serverExtended access963Repository stars

An MCP server that runs every reasoning step of your agent past six Agno specialists and merges their views into one answer

MCP serverExtended access306Repository stars

A CLI for any MCP server: list and call tools, resources and prompts, run mock and proxy modes, guard access and manage agent configs

CLIExtended access1.6KRepository stars

brainstorm-mcp runs debates among GPT, Gemini, DeepSeek, Claude and Ollama models in your editor and returns a three-point verdict

MCP serverExtended access73Repository stars

An MCP server that turns a request into a step plan with dependencies and keeps context across agent sessions

MCP serverExtended access2.2KRepository stars
Official

An MCP server for long-term agent memory: compresses dialogues into atomic facts, resolves pronouns and anchors relative time to absolute dates

MCP serverExtended access3.9KRepository stars

README in short

The README describes an adaptation of the Sequential Thinking server into a lightweight MCP server for recording reasoning steps. It has tools to record a thought, read session history and clear history, plus a prompt on honest use: it is a scratchpad and validator, not an external mind. It documents defensive filtering of untrusted text and options such as MAX_HISTORY_SIZE. Install is via npx, with a config for Claude Desktop and compatible clients.

README badge

Report a problem