constitution

What problem does it solve? AI agents working on long tasks face rule conflicts, novel situations, and ethical gray areas with no clear authority to resolve them. This Skill provides a foundational constitution — four ordered principles (Truth over Harmony, Evidence over Authority, Halt over Assumption, Outcomes over Process) — that governs every agent run and resolves conflicts when documented rules collide. ## Core Features & Use Cases - Always-On Law: Four priority-ordered principles that compose directly into a system-prompt core, so every agent decision inherits the same ethical floor. - Precedence Hierarchy: A ranked authority order (rules.md, constitution, engineering, ambient, documentation, domain skills) that deterministically resolves cross-file rule conflicts. - Reference Diagnostics: In-depth material on entropy accelerators (hollow output, purpose drift, silent complicity), convergence configurations, depth-of-analysis layers, and principled resistance for adjudicating novel or ethical situations. - Use Case: When an agent detects that a requested process step no longer serves the project goal, it applies Outcomes over Process — surfacing the concern with evidence instead of silently complying. ## Quick Start Invoke the constitution skill whenever rules conflict, an ethical call is needed, or a situation has no specific governing rule, and apply its four ordered principles to decide.