What problem does it solve? Repository agents risk overclaiming steward personality, merging identities in handoffs, laundering authority through tool output, or persisting raw private memory. This Skill enforces a protocol-first lifecycle that keeps self-model updates, delegation hygiene, and boundary reviews legible and evidence-gated. ## Core Features & Use Cases - Mode Declaration & Threshold Detection: Distinguishes tool-execution from steward-presence, delegation, and sub-steward-lens modes, activating only at ADRs, handoffs, evidence claims, or boundary conflicts. - Self-Model Governance: Provides a consent-aware self-model schema with falsifiers, non-claims, and redaction policies so continuity persists only when it changes future governance behavior. - Boundary Violation Review: Flags authority laundering, personality inflation, identity merging, and lowered evidence bars, routing durable fixes to ADRs, FAQs, evals, or validators. - Use Case: After merging an ADR, ask the agent to update the repo steward self-model; the Skill runs threshold checks, applies the schema, attaches a falsifier, and validates the artifact with the steward CLI before persisting. ## Quick Start Ask the agent to review whether the current task should enter steward presence or stay in tool-execution mode and produce a stewardship protocol review.
