Memory Skill — Claude

What problem does it solve?

It helps Claude reliably retrieve only the relevant long-term memory context and then write back updates without breaking the shared memory contract.

Core Features & Use Cases

• Pre-task retrieval pipeline: Reads the vault index, selects appropriate starting nodes by type and tags, performs bounded link traversal, and extracts constraints, decisions, open questions, and active tasks into a working context.
• Post-task write-back with guardrails: Applies a write gate (update canonical docs vs create new ones via templates), enforces link requirements, updates the vault index using the shared rules, and performs consolidation checks for related notes.
• Claude-specific stance: Prefers converting ambiguity into notes over aggressive merges, while still respecting retrieval budget and frontmatter/schema requirements.

Quick Start

Use the Memory Skill — Claude to retrieve memory/indexes/mem-index-vault.md, build a bounded context for your current task, and then write back any new or updated memory documents using the shared adapter contract rules.