Atom of Thoughts

Atom of Thoughts (AoT) MCP is a server that decomposes complex problems into independent atomic units of thought, using the dependencies between these units to deliver more robust reasoning and validated insights.
  • javascript

57

GitHub Stars

javascript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "kbsooo-mcp_atom_of_thoughts": {
      "command": "node",
      "args": [
        "/ABSOLUTE/PATH/TO/PARENT/FOLDER/atom-of-thoughts/build/index.js"
      ]
    }
  }
}

You can run Atom of Thoughts as an MCP server to structure complex reasoning into reusable atoms, verify hypotheses through a formal decomposition-contraction flow, and obtain reliable conclusions with configurable depth and automatic termination. This guide shows you how to use, install, and configure the AoT MCP server for practical, reliable reasoning tasks.

How to use

You use the AoT MCP server by registering it in your MCP client settings and then invoking its reasoning workflow through the client’s standard request interface. You start with premise atoms, build reasoning atoms from those premises, propose hypotheses, verify those hypotheses with verification atoms, and finally derive conclusions from verified hypotheses. When you need deeper exploration, you can initiate decomposition of atoms to generate sub-atoms and then contract them back after verification. The system automatically terminates when it reaches the maximum depth or when a high-confidence conclusion is found, and you can query for the best conclusion at any time.

How to install

npx -y @smithery/cli install @kbsooo/mcp_atom_of_thoughts --client claude

}]} ,{

## Configuration and runtime details

To run Atom of Thoughts as an MCP server locally, you configure a stdio-based MCP entry that launches a Node.js process running the built AoT server. The typical setup points the client to a local script and does not rely on a remote URL. The example below shows how to reference the built index.js file inside your local project and run it with Node.

{ "mcpServers": { "atom-of-thoughts": { "command": "node", "args": ["/ABSOLUTE/PATH/TO/PARENT/FOLDER/atom-of-thoughts/build/index.js"], "disabled": false, "autoApprove": [] } } }


Replace `/ABSOLUTE/PATH/TO/PARENT/FOLDER` with the actual absolute path to the project on your system. After saving the configuration, restart Claude Desktop or Cline to use the Atom of Thoughts MCP server.

Tools and capabilities

Atom of Thoughts provides distinct tools and endpoints to manage decomposition, verification, and conclusion extraction. Key capabilities include initiating decomposition of a given atom, adding sub-atoms to a decomposition, completing a decomposition, querying termination status, retrieving the best verified conclusion, and adjusting the maximum allowed decomposition depth.

Advanced usage notes

AoT uses five atom types to structure reasoning: premise, reasoning, hypothesis, verification, and conclusion. You track dependencies between atoms, assign a confidence score from 0 to 1 for each atom, and determine when an atom is verified. Use decomposition to explore deeper when needed, and contract back to derive high-confidence conclusions.

Troubleshooting and tips

If the AoT MCP server does not appear in the client, verify that the mcpServers entry is loaded by your MCP client and that the Node process specified in command is executable. Ensure the path to the built index.js is correct and that you restarted the client after adding the server. If you need faster results, consider using AoT-light with a smaller maximum depth and simplified verification.

Notes on deployment

When deploying, keep the AoT server on a stable Node.js runtime and ensure your environment allows the process to read and write any intermediate atom data. Document the atom IDs and their dependencies for auditability, and monitor the automatic termination behavior to avoid long-running sessions in time-sensitive scenarios.

Security considerations

Limit access to the MCP endpoint to trusted clients. Use secure channels if you expose the server remotely, and manage permissions to prevent unauthorized creation or modification of atoms, especially verification and conclusion atoms.

Available tools

startDecomposition

Begin decomposing a specified atom into smaller sub-atoms to enable deeper analysis.

addToDecomposition

Add a sub-atom to an ongoing decomposition using the decompositionId and atomId.

completeDecomposition

Finish the decomposition process, allowing the system to contract verified sub-atoms back into the parent atom.

termination_status

Check whether the current AoT process has reached a termination condition and obtain the reason.

best_conclusion

Retrieve the verified conclusion with the highest confidence score.

set_max_depth

Adjust the maximum allowed depth for decomposition to control analysis granularity.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational