sequential-thinking

Sequential reasoning for codebase analysis automates structured cognition by enforcing a clear trail of thoughts. It helps you define questions, record thoughts, and create branches for parallel analysis to build mental models before implementing changes.

Problem decomposition is best handled by extracting architecture patterns, data flows, and dependencies through structured reasoning. You can create parallel branches to compare approaches, then merge insights and store decisions with rationale to guide future work.

To start a thinking session for pattern extraction, initiate a new session, add an initial thought, create parallel branches for different analytical approaches, then merge insights back into the main thread to synthesize your findings.

You can manage parallel reasoning chains by creating branches for different approaches during codebase analysis. This allows you to compare patterns and data flows side-by-side before merging insights back into the main thread.

Yes, structured cognition requires explicit frontmatter metadata in your SKILL.md file with name and description fields. It also supports optional directories for scripts, references, and assets to organize your analysis.

Sequential reasoning ensures a clear trail of thoughts before implementation, preventing blind changes. By mapping components, interfaces, and constraints first, you build an accurate mental model that informs safer and more effective design decisions.