What problem does it solve? Decisions and solutions from past AI coding sessions often live only in old conversation logs, forcing you to re-ask questions, re-debug familiar bugs, or lose the rationale behind earlier choices. This Skill teaches the AI how and when to query the trellis mem CLI to recover that cross-session context. ## Core Features & Use Cases - Session Search & Extraction: Search indexed Claude Code, Codex, Pi Agent, and ZCode conversation logs by keyword, then extract cleaned dialogue filtered by phase or grep terms. - Phase-Aware Slicing: Cut sessions at Trellis task boundaries (--phase brainstorm / --phase implement) to recover planning discussions or execution loops separately. - Judgment-Driven Recall: Guidance on when reaching for memory is right (familiar bugs, decision retrieval, cross-session continuation) and when it is noise, plus how to act on findings (quote inline, update PRD, append notes, or just absorb). - Use Case: You hit a timeout bug that feels familiar. The AI runs trellis mem search "timeout" --global, finds the session where it was solved, and quotes the exact fix with a session-id citation. ## Quick Start Ask the AI to check whether you have discussed or solved a similar problem in past sessions before answering your current question.
