What problem does it solve? After context summarization or session handoff, an AI agent can lose track of canonical file paths, prior decisions, and the big-picture governance context, leading to repeated mistakes and orphaned work. This Skill enforces a structured recovery loop that searches transcripts and disk authority indexes before acting, so lost links and philosophical foundations are restored first. ## Core Features & Use Cases - Mandatory recovery loop: An 8-step checklist (pause, transcript search, disk search, classify, prove, wire, reply, verify) triggered on session start, handoff, or user signals like "big picture" or "lost link". - Truth-tree validation: Down-only checks from thorn (P0) to leaf (P4-6) with proof scripts such as authority coverage audits and law-purity validators. - FOUND reply format: A standardized human-comfort response naming the canonical path, the prior mistake, the proof, and what it enables. - Use Case: After a Cursor session is summarized mid-conversation, run the full recovery loop to relocate a lost LOCKED governance document via transcript anchors and the authority index, then reply with the FOUND format before continuing work. ## Quick Start Ask the agent to run the conscious recovery loop to find the lost thread link and big-picture context before continuing.
