Skip to main content
Glama
stabgan

Steelmind MCP

by stabgan
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

Steelmind MCP — Structured Thinking & Verification for AI Agents

npm version Docker License: MIT

The research-grounded reasoning MCP server for AI agents. Combines step-by-step sequential thinking with steel-manning verification — backed by 43+ cognitive science and AI research papers.

Steelmind gives your AI agent two tools:

  • think — Record structured reasoning steps with sequential decomposition. Embeds Socratic self-questioning and Polya's problem-solving method.

  • verify — Challenge conclusions with steel-manning before committing. Embeds dialectical evaluation from MetaCrit and SIEV research.

The code is minimal. The descriptions do the heavy lifting — tool descriptions account for ~80% of reasoning improvement per Anthropic τ-bench research.

Why Steelmind?

Feature

Think MCP

Sequential Thinking

Steelmind

Step tracking

✗

✓

✓

Adjustable step count

✗

✓

✓

Cognitive mode separation

✗

✗

✓

Steel-manning verification

✗

✗

✓

Socratic self-questioning

✗

✗

✓

Research-grounded descriptions

✗

✗

✓

Verify nudge on completion

✗

✗

✓

Tool count

1

1

2

Key research insight: MetaCrit (arxiv 2507.15015) proved that separating reasoning generation from reasoning evaluation prevents self-bias and improves accuracy by up to 76%. Sequential-thinking uses one tool for both. Steelmind separates them.

Related MCP server: Sequential Thinking MVP Server

Quick Start

npx (no install)

{
  "mcpServers": {
    "steelmind": {
      "command": "npx",
      "args": ["-y", "@stabgan/steelmind-mcp"]
    }
  }
}

Docker

{
  "mcpServers": {
    "steelmind": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "stabgan/steelmind-mcp"]
    }
  }
}

npm global install

npm install -g @stabgan/steelmind-mcp
{
  "mcpServers": {
    "steelmind": {
      "command": "steelmind-mcp"
    }
  }
}

How It Works

The think tool

Records a structured reasoning step with sequential tracking.

Input:

{
  "thought": "What are the dependencies? Need to check imports before refactoring.",
  "thoughtNumber": 1,
  "totalThoughts": 3,
  "nextThoughtNeeded": true
}

Output (mid-sequence):

[Thinking 1/3]

What are the dependencies? Need to check imports before refactoring.

Output (final step — includes verify nudge):

[Thinking 3/3]

My conclusion: use the adapter pattern for backward compatibility.

---
Thinking complete. Before acting on this conclusion, use the verify tool to challenge it.

The verify nudge appears in the tool result (not just the description), making it far more likely the model will actually call verify. Tool results get different attention treatment than descriptions — they're processed as fresh context.

The verify tool

Challenges your reasoning with steel-manning before you commit.

Input:

{
  "concern": "The adapter pattern adds complexity. Is the simpler approach actually better?"
}

Output:

The adapter pattern adds complexity. Is the simpler approach actually better?

Pure identity function — returns your concern unchanged. The value is in the description, which prompts: "Steel-man the opposition: What is the strongest argument that your conclusion is wrong?"

The workflow

think(step 1/3) → think(step 2/3) → think(step 3/3) → [verify nudge] → verify → act
                                          ↑
                                  adjust totalThoughts if needed

Research Foundation

Steelmind's design is grounded in 43+ research papers. Key findings:

Paper

Finding

How Steelmind Uses It

MetaCrit (arxiv 2507.15015)

Separating generation from evaluation prevents self-bias

Two separate tools: think (generate) + verify (evaluate)

Anthropic τ-bench

Optimized tool descriptions yield 54% improvement

Descriptions are the primary scaffold, not code

Think2 (arxiv 2602.18806)

Structured metacognition yields 3x self-correction

Sequential step tracking + Socratic questioning

SIEV (ICML)

Models lose 40+ points under dialectical evaluation

Steel-manning prompt in verify description

Scaling TTC (arxiv 2408.03314)

Difficulty-adaptive compute improves efficiency 4x

Adjustable totalThoughts

EasyTool (NAACL 2025)

Concise descriptions outperform verbose ones

~100 word descriptions

ToolACE

"When NOT to use" improves irrelevance detection 6→84%

Negative guidance in both descriptions

Cognitive Foundations (arxiv 2511.16660)

External scaffolding improves performance up to 72%

Research-grounded cognitive frameworks

Compatible Clients

Works with any MCP-compatible client:

  • Claude Desktop / Claude Code

  • Cursor

  • Windsurf

  • Kiro

  • Cline

  • Any client supporting MCP stdio transport

Compatible Models

Designed for frontier models but works across families:

  • Claude (Opus, Sonnet) — native MCP

  • GPT-5 / GPT-4o / o-series — via MCP adapters

  • Gemini — via MCP adapters

  • DeepSeek — via MCP adapters

Development

npm install          # Install dependencies
npm run build        # Compile TypeScript
npm test             # Run 90 tests
npm run lint         # ESLint
npm run format       # Prettier
npm start            # Run the server

License

MIT

Available Tools

2 tools
thinkA

Use this tool to record a structured reasoning step. It will not obtain new information or change any state — it appends your thought to the log. Use it when you need to: process results from previous tool calls before acting, plan your approach to a multi-step task, analyze a complex situation before deciding, or navigate environments with detailed policies. Do NOT use for simple single-step tasks or restating without analysis. When thinking, ask yourself: What am I assuming? What evidence supports this? What's my plan, and what could go wrong? You can adjust totalThoughts up or down as your understanding deepens. When you set nextThoughtNeeded to false, use the verify tool to challenge your conclusion before acting.

ParametersJSON Schema
NameRequiredDescriptionDefault
thoughtYesYour current thinking step.
thoughtNumberYesCurrent thought number in the sequence.
totalThoughtsYesEstimated total thoughts needed. Can be adjusted up or down as you progress.
nextThoughtNeededYesWhether another thinking step is needed after this one.

TDQS

A4.6/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Clearly states the tool is non-destructive and read-only: 'It will not obtain new information or change any state — it appends your thought to the log.' This is fully transparent about its behavior.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is longer than necessary, including some meta-cognitive advice that could be seen as extraneous. While front-loaded, it could be more concise.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's simplicity, the description covers the main aspects: purpose, usage, and behavior. It also references the sibling tool and provides procedural tips, making it fairly complete.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema has 100% coverage, so baseline is 3. The description adds value by explaining how to use parameters like adjusting totalThoughts and using verify when nextThoughtNeeded is false.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: 'record a structured reasoning step.' It specifies it does not obtain new information or change state, and it distinguishes itself from the sibling tool 'verify' by explaining when to use each.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides explicit guidance on when to use (processing results, planning, analyzing, navigating policies) and when not to use (simple single-step tasks, restating without analysis). Also instructs to use the verify tool after the final thought.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

verifyA

Use this tool to challenge and evaluate your reasoning before committing to an action. It will not obtain new information or change any state — it logs your critical self-assessment. Use it when you need to: check if your planned action complies with all requirements, validate reasoning before committing, assess edge cases, or evaluate tool results for correctness. Do NOT use to confirm what you are already confident about. When verifying, steel-man the opposition: What is the strongest argument that your conclusion is wrong? If you can't defeat it, reconsider. If your verification reveals a flaw, use the think tool to revise your approach.

ParametersJSON Schema
NameRequiredDescriptionDefault
concernYesYour critical assessment or concern to verify.

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided, so description fully discloses behavior: it only logs self-assessment, obtains no new info, and changes no state. This is complete transparency.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is well-structured with purpose first, then usage, then how-to. While slightly lengthy, every sentence adds value and there is no fluff.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the simple tool (one param, no output schema, no annotations), the description is thorough. It covers purpose, behavior, usage guidelines, and verification method, leaving no gaps.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% for the single 'concern' parameter, providing baseline 3. The description adds context by explaining that the parameter holds a critical assessment or concern, which is consistent with the tool's purpose.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: to challenge and evaluate reasoning before committing to an action. It specifies that it does not obtain new information or change state, distinguishing it from its sibling 'think'.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicitly provides when to use (check compliance, validate reasoning, assess edge cases, evaluate results) and when not to use (confident cases). Includes guidance on how to verify by steel-manning the opposition.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 2 tool updatesv2.0.1
    • First observedthink
    • First observedverify
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.7/5.0

Scored across 2 tools

Disambiguation5/5

Both tools have clearly distinct purposes: 'think' is for recording reasoning steps, while 'verify' is for challenging and evaluating reasoning before action. There is no ambiguity between them.

Naming Consistency5/5

Tool names follow a consistent single-word verb pattern ('think' and 'verify'), which is appropriate for metacognitive tools and perfectly consistent.

Tool Count4/5

With only two tools, the set is minimal but well-scoped for a server focused on reasoning and self-reflection. It earns its place without being over or under.

Completeness5/5

The tools cover the essential metacognitive cycle of thinking and verifying. There are no obvious gaps for the server's intended purpose of enhancing reasoning.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers