This repository uses “System 2” as a practical metaphor for slow, deliberate review. It does not claim that a language model literally switches into Daniel Kahneman’s human cognitive system.
Useful interpretation
A System-2-style agent workflow means:
- avoid instant closure;
- inspect before acting;
- split the task into smaller problems;
- compare routes;
- test the selected route;
- disclose failed or missing checks;
- trace claims to evidence.
Unsafe interpretation
Do not say:
This prompt tricks the AI into its most advanced mathematical pathway.Say instead:
This prompt uses deliberate-work controls that make the agent produce reviewable evidence before confidence.Copy-ready control clause
Use deliberate-work mode: do not answer from first impression. Inspect inputs, decompose the task, compare at least one alternative when risk is material, act minimally, verify, and report limits before final status.Required research vocabulary cross-reference
This document participates in the same control vocabulary used across the library: CoT-safe public reasoning, Tree-of-Thoughts branch evaluation, ReAct observation loops, Least-to-most decomposition, Self-consistency checks, Self-refinement after failed checks, Process supervision, and traceability.