From memory-mason
Queries Obsidian vault knowledge base via session context, index, atlas, concepts, and synthesis pages. Answers questions on past decisions, patterns, lessons, or technical knowledge from AI conversations with [[wikilink]] citations.
npx claudepluginhub s-gryt/memory-mason --plugin memory-mason[question]This skill is limited to using the following tools:
Answer a knowledge-base question using index-guided retrieval and cite sources with [[wikilinks]].
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Guides code writing, review, and refactoring with Karpathy-inspired rules to avoid overcomplication, ensure simplicity, surgical changes, and verifiable success criteria.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Share bugs, ideas, or general feedback.
Answer a knowledge-base question using index-guided retrieval and cite sources with [[wikilinks]].
This command is operational only. Do not write /mmq, /memory-mason:mmq, or their execution chatter back into the vault.
Before any other reasoning, resolve vault config in this priority order:
./.env./memory-mason.json~/.memory-mason/.env~/.memory-mason/config.jsonResolve:
Use the source that provides the vault path.
Subfolder rules:
.env file, use MEMORY_MASON_SUBFOLDER from that same file when present, otherwise default to ai-knowledge.memory-mason.json or ~/.memory-mason/config.json, use its subfolder.Do not claim config is missing until you have attempted all four locations above. If none provide a vault path, state that the knowledge base is not initialized because no supported Memory Mason config source was found.
Use these paths:
context.md as a fast cache of recent context, not a replacement for the knowledge base.context.md directly answers the question with enough precision, answer immediately.context.md is relevant but incomplete, extract likely page names, tags, dates, and follow-up questions from it, then continue.context.md when they are relevant._raw/YYYY-MM-DD/ sources were compiled._raw/YYYY-MM-DD/NNN.md chunks directly instead of assuming the vault is current.context.md only as retrieval assistance. Do not cite _meta/context.md as authoritative when a durable page exists./mmc if relevant information appears to exist in _raw/ but is not compiled into durable pages yet._meta/context.md to accelerate retrieval, not to skip article reads when the answer requires durable citations.