What problem does it solve?
Provides an explicit, repeatable methodology for multi-step reasoning so models and users can produce verifiable intermediate results, handle branching and revisions, and avoid silent or circular reasoning. It replaces a legacy MCP tool by encoding step numbering, revision checkpoints, branch exploration, dynamic scope updates, and final verification into plain prompting.
Core Features & Use Cases
- Numbered Step Execution: Work through 3–10+ numbered thoughts with a clear sub-goal, reasoning, and intermediate conclusion for each step.
- Revision Checkpoints: Mandatory self-checks every few steps with explicit "Revising Step N" markers to correct prior conclusions and list downstream impacts.
- Branch Exploration: Visible fork points with labeled approaches, short comparative evaluation, and a commitment rationale; briefly explore alternatives when consequential.
- Dynamic Scope Adjustment & Verification: Update estimated step counts mid-analysis and conclude with a verification checklist that traces requirements to specific steps.
- Use Cases: Deep debugging sessions, architectural decision-making, formal proofs, and any analysis needing stepwise traceability and error-correction.
Quick Start
Apply the sequential-thinking protocol to diagnose intermittent service failures by decomposing the problem into numbered steps, running periodic revision checkpoints, exploring up to two branches when necessary, and finishing with a verification checklist.
