critical-thinker

Pressure-testing architecture design challenges high-impact choices by comparing alternatives and surfacing objections. It uses a subagent to read project memory, evaluate options, and log the final rationale to DECISIONS.md.

Evaluating database schema and API design decisions requires a subagent-driven critique that reads project memory files, compares alternatives, and surfaces the strongest objections before logging the outcome to DECISIONS.md.

You need a structured critique workflow for decision analysis when a choice affects three or more future gates, such as project starts, agent architecture, stack selection, or database schema design.

Yes, a subagent-driven critique process compares a single-agent approach against complex orchestration by reading project context, evaluating alternatives, and exposing whether a simpler design reduces risk before the decision becomes load-bearing.

Decision logging preserves architecture review rationale across context resets by recording the final outcome and reasoning in DECISIONS.md, ensuring the critique process survives future gates and project memory resets.

Slowing down major product decisions prevents expensive changes by exposing hidden complexity, missing evidence, and simpler alternatives. The critique workflow ensures choices are fully evaluated before they affect multiple future gates.