"Understanding Claude AI's Reasoning and Thinking Ability" matching MCP tools:

• ask_claudeA

  claude-max-mcpA

  Delegate complex tasks to Claude for code generation, reasoning, and engineering. Include all context in the prompt, as Claude has no conversation memory.

  MIT

• ultrathinkA

  UltraThinkA

  Analyze complex problems through sequential thinking, track assumptions, and manage multi-step reasoning with confidence scoring and session-based context.

  MIT

• reviseA

  MAXential Thinking MCPA

  Replace a previous thought with updated reasoning, preserving the original in revision history. Corrects incomplete or wrong thinking.

  MIT

• configure_tracingA

  ask-fableA

  Toggle ask-fable reasoning-trace capture and live thinking at runtime. Persist trace_mode (full/safe) and stream_reasoning settings without editing config files or restarting.

  MIT

• crashA

  CRASH - Cascaded Reasoning with Adaptive Step HandlingA

  Break down complex problems into structured reasoning steps to track thinking, expected outcomes, and next actions for systematic problem-solving.

  MIT

• cph_thinking_summaryA

  Claude Project History MCP ServerA

  Infers thinking-time breakdown for recent turns by measuring gaps between tool calls. Returns per-turn and aggregate stats to analyze Claude Code's extended thinking.

  MIT

• jev_modulate_reasoning_effortA

  jev-harnessA

  Calibrate reasoning effort and generation lease steps for the next LLM call, mapping provider parameters to prevent reasoning token waste.

  MIT

• classify_complexityA

  ThinkGateA

  Classify prompt complexity to select the optimal model tier and thinking effort, returning tier, effort, suggested model, confidence, and reasoning.

  MIT

• list_modelsA

  cursor-agent-mcpA

  Lists available model identifiers for the --model parameter, showing options like gpt-5, sonnet-4-thinking, and claude-opus-4-8.

  MIT

• get_task_outputA

  ClaudeArmyA

  Retrieve the full output from a Claude Code agent task to see its reasoning, actions, and completed work.

  MIT

• darbot_deepmindA

  Darbot Deepmind MCP ServerA

  Analyze complex problems through structured thinking that adapts and evolves as understanding deepens, enabling dynamic problem-solving with revision and multi-step reasoning.

  MIT

• extract_reasoning_chainA

  mcp-agent-trace-inspectorA

  Extract reasoning and thinking steps from a trace by matching tool names or content. Use a trace_id to retrieve only the reasoning chain.

  MIT

• reasonB

  DeepSeek-Claude MCP ServerB

  Process queries using DeepSeek's R1 reasoning engine to generate structured analysis for Claude integration, handling complex multi-step reasoning tasks with precision.

  MIT

• brainstormB

  ccg-mcp-toolB

  Generate novel ideas by applying creative frameworks (SCAMPER, Design Thinking) with domain context, idea clustering, feasibility analysis, and iterative refinement. Supports Gemini, Codex, and Claude.

  MIT

• getHeroAbilitiesB

  Marvel Rivals MCPA

  Fetch detailed ability kits for Marvel Rivals heroes to understand mechanics, cooldowns, and combat strategies. Use this tool to access primary fire, abilities, ultimates, and passive information for specific heroes.

  MIT

• wisdom_clarifyB

  mcp-wisdomA

  Clarify unclear thinking by applying Socratic questioning to examine assumptions and understand concepts better.

  MIT

• sequential_thinkingC

  Sequential Thinking MVP ServerB

  Break down complex problems into manageable steps for structured analysis. Track reasoning progress, revise previous steps, and explore alternative paths to solve challenging problems systematically.

  -

• deep_thinkingA

  win-cli-mcp-tmyyB

  Break down complex problems into manageable steps, revise reasoning as understanding deepens, and explore alternative paths through structured sequential thinking.

  -

• cognitive_reasoningA

  Advanced Cognitive MCP ServerA

  Process complex problems through multi-modal reasoning, collaborative AI personas, and dynamic knowledge graphs. Supports analytical, creative, and critical thinking modes to generate well-reasoned solutions.

  -

• ollama_generate_textA

  Claude SidekickA

  Generate simple text locally with Ollama for error messages, placeholder content, boilerplate code, and routine documentation, saving Claude tokens for complex reasoning.

  -

• classify_effortA

  kenpomB

  Determines the appropriate reasoning depth by classifying query effort levels into low, medium, or high, with model hints and thinking budget for efficient task execution.

  -