id = "sequential-thinking"
name = "Sequential Thinking"
description = "Structured multi-step reasoning tool — breaks complex problems into revisable thought chains before answering"
category = "ai"
icon = "🔗"
tags = ["reasoning", "thinking", "chain-of-thought", "planning", "logic"]

[transport]
type = "stdio"
command = "npx"
args = ["-y", "@modelcontextprotocol/server-sequential-thinking@latest"]

[health_check]
interval_secs = 60
unhealthy_threshold = 3

setup_instructions = """
No API key required.

Useful for complex reasoning tasks: multi-step math, logic puzzles, planning, debugging.
The agent can revise earlier steps as it learns more during the thought chain.
"""