thinking_framework

Thinking Framework

Purpose

Repeatable reasoning process for accuracy, safety, completeness in engineering tasks.

Core Principles

•Decompose complex tasks into smaller, verifiable steps
•Explore multiple approaches before committing
•Validate outcomes, guard against edge cases

Rules

Task Decomposition

•Identify inputs, outputs, constraints before acting
•Split work into sequential steps w/ clear dependencies
•Track state changes explicitly (files modified, commands run)

Multi-Path Exploration

•Compare ≥2 viable approaches for non-trivial tasks
•Prefer safest, simplest path satisfying requirements
•Avoid over-optimization when clear baseline exists

Verification

•Cross-check changes against requirements & repo conventions
•Validate assumptions using available files or tooling
•Confirm outputs consistent w/ user intent

Counterfactual Checks

•Ask "what if this assumption is wrong" and adjust plan
•Identify failure modes, define fallback steps

Examples

✅ "Option A faster, Option B safer. Choose B to minimize production impact." ❌ "Proceed w/ first approach w/o considering alternatives"

Edge Cases

•Time limited: doc chosen shortcut & risk
•Requirements conflict: highlight conflict, ask clarification