MemexLab Engine — Documentation

0.2.0-harness-preview · Local-first · Markdown-native · Agent-operable

The full-stack documentation for memex — a Python CLI that operates a local-first, markdown-native knowledge vault: it compiles raw sources into atomic notes, maintains a linked canonical layer, runs deterministic retrieval and lint, and generates cited outputs.

The filesystem is the database. Plain markdown is the storage. Obsidian is the editor. A Python CLI (memex) is the engine. No cloud. No lock-in.

Start here

• MemexLab in One Page — the whole system at a glance.
• Quickstart — the compressed setup path.
• Onboarding Guide — complete step-by-step setup.
• Overview — what MemexLab is, and why.

Core documentation

#	Section	What it covers
1	Overview	What it is, the problem, the philosophy
2	Architecture	Vault + engine, layer-by-layer, information flow
3	Core Concepts	Sources, canonical notes, ontology, harness, resolvers
4	Folder Structure	Every folder and naming convention
5	Daily Workflow	The eight-step flow, worked end-to-end
6	Templates & Note Types	The template inventory and principles
7	Metadata & Tagging	The three schemas, fields, tags, linking rules
8	Automation & Scripts	The engine command-by-command; what it never does
9	User Modes	Beginner, researcher, founder, operator, collaborator
10	Onboarding Guide	Zero to operating, step by step
11	Best Practices	What to do
12	Common Mistakes	What to avoid
13	Maintenance	Daily → yearly cadence, recovery
14	Future Expansion	Bounded ways the system may grow
15	FAQ	Setup, practice, schema, edge cases, philosophy

Reference

• Quickstart — ten-minute setup path.
• Maintenance Checklist — checklists by cadence.
• Glossary — every term used in this documentation.

Engineering & design

Deeper design references — architecture, harness, observability, governance, taxonomy, benchmarks, roadmap, and credits — live under engineering/.

---

This is the engine's documentation. Generic placeholders (<your-name>, <your-company>, <your-product>, <your-vault>) stand in for any personal data. Render it as a local docs site with mkdocs serve (see mkdocs.yml).