CRASH

MCP server for structured and efficient reasoning with step validation, branching, and revisions.
  • typescript

67

GitHub Stars

typescript

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": {
    "nikkoxgonzales-crash-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "crash-mcp"
      ],
      "env": {
        "CRASH_NO_COLOR": "false",
        "MAX_HISTORY_SIZE": "100",
        "CRASH_STRICT_MODE": "false",
        "CRASH_OUTPUT_FORMAT": "console",
        "CRASH_ENABLE_SESSIONS": "false",
        "CRASH_SESSION_TIMEOUT": "60",
        "CRASH_MAX_BRANCH_DEPTH": "5"
      }
    }
  }
}

CRASH MCP Server lets you break down complex problems into trackable steps with confidence tracking, revision support, and branching for exploring alternatives. It streamlines iterative reasoning in a structured way, helping you manage multi-step tasks efficiently and safely.

How to use

You use CRASH with an MCP client to drive iterative reasoning sessions. Start a session, then add steps that describe each analysis, decision, and action. Track confidence, mark revisions when you correct course, and create branches to compare alternative approaches. You can output results in consoles, JSON, or Markdown, so you can review, share, or export your reasoning history.

How to install

Prerequisites: you need Node.js installed on your system. You can verify by running node -v and npm -v in your terminal. If Node.js is not present, install it from the official Node.js website or via your system package manager.

Install CRASH MCP globally or run it directly with a package manager. Use one of the following approaches depending on your preference.

npm install crash-mcp
npx crash-mcp

Configuration and quick setup

Most MCP clients use a simple JSON configuration to connect to the CRASH MCP server. The basic setup connects via a local runtime when you run the MCP from your terminal.

{
  "mcpServers": {
    "crash": {
      "command": "npx",
      "args": ["-y", "crash-mcp"]
    }
  }
}

Windows users and alternative runtimes

If you are on Windows, you can use a CMD wrapper to run CRASH with the same capabilities.

{
  "mcpServers": {
    "crash": {
      "command": "cmd",
      "args": ["/c", "npx", "-y", "crash-mcp"]
    }
  }
}

Using environment variables

You can tailor CRASH behavior with environment variables. The following options are commonly used to control strict validation, session timeouts, output format, and more.

{
  "mcpServers": {
    "crash": {
      "command": "npx",
      "args": ["-y", "crash-mcp"],
      "env": {
        "CRASH_STRICT_MODE": "false",
        "MAX_HISTORY_SIZE": "100",
        "CRASH_OUTPUT_FORMAT": "console",
        "CRASH_SESSION_TIMEOUT": "60",
        "CRASH_MAX_BRANCH_DEPTH": "5"
      }
    }
  }
}

Using Docker

You can run CRASH in a Docker container to keep your environment clean.

FROM node:18-alpine
WORKDIR /app
RUN npm install -g crash-mcp
CMD ["crash-mcp"]
{
  "mcpServers": {
    "crash": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "crash-mcp"]
    }
  }
}

Alternative runtimes

If you prefer other runtimes, you can use Bun or Deno to run CRASH.

{ "command": "bunx", "args": ["-y", "crash-mcp"] }
{ "command": "deno", "args": ["run", "--allow-env", "--allow-net", "npm:crash-mcp"] }

Available tools

structured_steps

Track thought process, outcomes, and next actions for each step.

confidence_tracking

Assign a 0-1 confidence score to each step and warn on low confidence.

revision_mechanism

Revise previous steps with original steps marked as revised and clearly labeled.

branching

Explore multiple solution paths with depth limits and branch identifiers.

dependency_validation

Declare and validate dependencies between steps to ensure correct ordering.

session_management

Group related reasoning chains into sessions with automatic timeout cleanup.

output_formats

Produce results in console, JSON, or Markdown formats.

validation_modes

Support strict and flexible validation modes depending on user needs.

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