first-principles-mode

What problem does it solve?

This skill eliminates shallow, surface-level answers to complex, ambiguous, or repeated-failure problems by running a deep, mechanism-level read-only analysis that tests competing explanations before offering any implementation judgment.

Core Features & Use Cases

• Adversarial Council Protocol: Runs a multi-lane adversarial review with structured rebuttal rounds to test competing explanations, surface hidden constraints, and avoid groupthink for high-leverage or ambiguous questions.
• Optional Research Escalation: Supports --deep-research for live web-backed checks of current operator practice and external sources, and --pro-analysis for ChatGPT Pro browser escalation to validate local findings against external expertise.
• Use Case Example: Use this skill when your system keeps collapsing under load, a project plan has hidden reasoning gaps, or you need to understand the real root cause of a repeated failure before implementing fixes.

Quick Start

Use the first-principles-mode skill to run a deep read-only analysis of your ambiguous hard problem or repeated system failure, with adversarial testing of competing explanations, before implementing any fixes.