What problem does it solve? Turning an AI memory architecture document into working production code is error-prone: wiring retrieval and write paths, enforcing per-user isolation, handling dedup and retention, and validating everything takes days of manual effort. This Skill automates that deployment end-to-end with checkpoints and tests. ## Core Features & Use Cases - 11-Phase Autonomous Build: Runs doc preflight, codebase comprehension, drift checking, stack research, prerequisites, setup, wiring, quality, hardening, eval, and a readiness report with resumable checkpoints logged to .memory-ship/build-log.md. - Framework-Native Wiring with Safety Wrapper: Delegates retrieval, ranking, dedup, and classification to the chosen memory framework (Mem0, Zep, pgvector, Pinecone, etc.) while enforcing universal guarantees like user_id isolation, prompt-injection sanitization, token budgets, and GDPR forgetUser. - Verified Quality Gates: Every code-touching phase ends with step tests covering dedup, contradiction supersession, retention, leak tests, and degraded-mode behavior before the build is declared done. - Use Case: After running /memory-design to produce an architecture doc for a support chatbot, invoke /memory-ship to install packages, create migrations, wire memory into every LLM call site, and receive a readiness report with eval results. ## Quick Start Type /memory-ship in your AI assistant after generating an architecture doc with /memory-design to build the memory system into your project.
