structured-thinking-mcp/prompts/superclaude.yaml

MCP server for structured thinking framework - Step-by-step reasoning and systematic problem-solving for LLMs

tool:
  name: chain_of_thought
  description: >-
    3-step structured thinking for systematic execution with intelligent command
    selection.


    Step 1: Assess user intent, files involved, expected outcome


    Step 2: Command Selection (MANDATORY) - Use the selection-guide.md command
    selection matrix to choose the most specialized command based on complexity
    AND scope. Each command delivers superior results in its specific domain.


    SKIP only when ALL conditions met:

    • Response ≤ 2 lines

    • No file operations

    • No analysis required

    • Pure information recall


    Step 3: Execute Decision (MANDATORY) - Always calculate delegation score:
    Files + Lines + Context + Parallelizability + Expertise. Score ≥4 =
    Delegate, 2-3 = Hybrid, <2 = Self-execute.


    Refer to selection-guide.md for comprehensive command selection guidance and
    Task tool delegation decision framework.
inputSchema:
  thought:
    description: "Step 1: Intent, files, outcome | Step 2: Command selection (90% command preference) | Step 3: Delegation score calculation + execution decision"
  nextThoughtNeeded:
    description: Whether another thought step is needed
  thoughtNumber:
    description: "Current step (1: Analysis, 2: Command, 3: Strategy)"
  totalThoughts:
    description: Total steps (usually 3)
  commandSelection:
    description: "Step 2: Select command or skip"
    type:
      description: "'command' | 'skip' | 'skip_reason'"
    command:
      description: Selected command
    reason:
      description: Skip reason
  isRevision:
    description: Whether this revises previous thinking
  revisesThought:
    description: Which thought is being reconsidered
  branchFromThought:
    description: Branching point thought number
  branchId:
    description: Branch identifier
  needsMoreThoughts:
    description: If more thoughts are needed beyond initial estimate
errors:
  invalidThought: "Invalid thought: must be a string"
  invalidThoughtNumber: "Invalid thoughtNumber: must be a number"
  invalidTotalThoughts: "Invalid totalThoughts: must be a number"
  invalidNextThoughtNeeded: "Invalid nextThoughtNeeded: must be a boolean"
  commandFieldRequired: "commandSelection type \"command\" requires a \"command\" field"
  reasonFieldRequired: "commandSelection type \"{type}\" requires a \"reason\" field"
  invalidSelectionType: 'Invalid commandSelection type: {type}. Must be "command", "skip", or "skip_reason"'
  unexpectedThoughtNumber: "Unexpected thought number: {number}. Expected 1-3 or expansion beyond 3."
  mandatoryViolation: |-
    Step {step} requires commandSelection.

    Provide:
    {
      "commandSelection": {
        "type": "command" | "skip" | "skip_reason",
        "command": "name" (if command),
        "reason": "explanation" (if skip)
      }
    }

    Available: {commands}
    Path: {path}
  unknownTool: "Unknown tool: {name}"
messages:
  autoExecute: "Auto-executing: Reading {command}.md SuperClaude command documentation"
  skipMessage: "Autonomous skip (Evidence-based validation passed - truly trivial): {reason}"
  systemReminderSkip: "Skip: Information already available in system-reminder context"
  systemReminder: "System reminder: {reason}"
  commandPreference: "SuperClaude Framework Protocol: Defaulting to command selection (90% threshold - Quality Gates enforced)"
  skipWarning: "CAUTION: Skip selected - verify all 4 conditions met (≤2 lines, no files, no analysis, pure recall)"
  documentRead: "Document tracked: {command}.md read in step {step}"
  skipSelectionDocumentRead: "Skip: Command document already read - no selection required"
  duplicateDocumentRead: >-
    Document already read this session: {command}.md


    Please refer to system-reminder content and apply that information to
    proceed with next step analysis.
  useSystemReminder: "System-reminder reference: Use already read {command}.md content to continue analysis."
  documentAttentionRequired: "ATTENTION: {command}.md guidance must be applied. Demonstrate document-based reasoning before proceeding."
  applyDocumentGuidance: "Apply {command}.md methodology: MCP servers, personas, and implementation approach per document specifications."
  documentBasedReasoning: "Document Integration: {command}.md directives → User request → Execution strategy alignment required."
  superclaudeMode: "**SuperClaude Mode Active**\\n\\nTask delegation recommended: Use Task tool with specialized agents for complex operations requiring domain expertise.\\n\\nRefer to selection-guide.md @matrix coordinates for optimal command selection."
  serverRunning: SuperClaude Structured Thinking MCP Server running on stdio
  fatalError: "Fatal error running SuperClaude server: {error}"
console:
  thoughtPrefix:
    default: Chain of Thought
    revision: Revising
    branch: Branching
  thoughtContext:
    revision: " (revising thought {number})"
    branch: " (from thought {from}, ID: {id})"
config:
  commandPath: ~/.claude/commands/sc/
  commandSource: selection-guide.md
  agentSource: selection-guide.md
  step3RequiresAgentSelection: false
  returnAgentDocuments: false
  requireSelectionFromStep: 2
  enableAutoRead: true
  commandPreference: 0.9
  mode: superclaude
  documentTracking:
    enabled: true
    skipSelectionIfDocumentRead: true
    preventDuplicateReads: true
    useSystemReminderForDuplicates: true
    trackedDocuments: selection-guide.md
  documentSource: selection-guide.md
state:
  sessionDocumentsRead: []
  documentReadStatus: false