Skip to main content
Glama

CRASH - Cascaded Reasoning with Adaptive Step Handling

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
CRASH_NO_COLORNoDisable colored outputfalse
MAX_HISTORY_SIZENoMaximum steps to retain100
CRASH_STRICT_MODENoEnable legacy validation rulesfalse
CRASH_OUTPUT_FORMATNoOutput display formatconsole

Schema

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Tools

Functions exposed to the LLM to take actions

NameDescription
crash

Facilitates structured, iterative reasoning for complex problem-solving and analysis. Use this to organize thoughts, break down tasks, and plan actions step-by-step. This tool is ideal for scenarios requiring careful consideration of multiple factors, such as code optimization, system design, or strategic planning.

When to use CRASH:

  • When user says "use crash" or similar phrases
  • For multi-step or complex tasks requiring careful breakdown
  • For understanding, planning, or analyzing codes or content
  • When systematic reasoning is needed
  • When you need to revise or correct previous reasoning
  • When exploring multiple solution paths

Key Features:

  • Flexible purpose types (analysis, action, validation, exploration, hypothesis, etc.)
  • Confidence tracking for uncertain situations
  • Revision mechanism to correct previous steps
  • Branching support for parallel exploration
  • Structured tool integration
  • No rigid formatting requirements (configurable)

Instructions:

  • Each step must specify a clear purpose and reasoning approach
  • Track context to avoid redundancy
  • Use confidence scores when uncertain (0-1 scale)
  • Revise previous steps when needed using revises_step
  • Branch to explore alternatives using branch_from
  • Continue until task completion or solution found

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/nikkoxgonzales/crash-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server