Your Obsidian vault is not a notebook. It's persistent memory for every AI agent you run.
Overview
Most developers treat Obsidian as a place to store notes. Navique treats it as the long-term memory layer for your entire AI workflow. Every agent session leaves breadcrumbs — claim-based notes with link-as-sentence connections that accumulate into a living knowledge graph. A CLAUDE.md file at the vault root acts as the agent's instruction set: tone, structure rules, naming conventions. Over weeks, the vault stops being a collection of files and becomes a structured thinking system that every agent can draw from the moment a session begins. You don't take notes. You operate a knowledge system.
AI Thinker Vault — Obsidian as Agent Memory
The vault is the agent's persistent memory. Each session reads prior context from the vault and writes new breadcrumbs back — claim-based notes named as assertions ('React Server Components eliminate client bundle for data-heavy routes') linked to other notes via full-sentence wikilinks. A CLAUDE.md at the vault root defines the agent's instruction set: how to name notes, when to link, what structure to follow, and what quality standard to maintain. The result is a knowledge base that compounds over weeks and months — not a pile of files, but a structured thinking system that gets smarter the more you use it.
Automated Vault Curation via Scheduled Hatch
A pre-configured Eggspert Hatch deploys a curation agent on a schedule you control. It processes the inbox folder, synthesises connections between new and existing notes, updates the vault index, merges duplicates, strengthens weak links, and fires a completion notification when done. You configure the schedule (daily, twice-daily, weekly), scope (full vault or specific folders), curation depth (light pass or deep synthesis), and quality standard. Set it and forget it — wake up to a vault that organised itself overnight.
Vault as Agent Context Source
Every Eggspert agent — not just the vault curation Hatch — can pull from the vault as a context source. When you launch a coding agent, a research agent, or a deployment agent, it starts the session with access to months of accumulated thinking: architectural decisions, debugging breadcrumbs, project context, meeting notes. This means agents are not starting from zero. They inherit institutional knowledge from day one of the session. The vault becomes shared memory across every agent you run.
Vault Browser with Full-Text Search
A complete Obsidian vault browser built into Navique. The note list is searchable by title and content. A folder tree allows navigation by directory. Note content is rendered with markdown formatting. Notes can be opened in their full detail sheet with forward/backlink exploration.
Quick Capture
A floating action button in the Vault view launches a Quick Capture sheet: enter a note title, optional tags, and content. The note is created directly in the connected Obsidian vault with appropriate metadata. No need to open Obsidian for a quick thought capture.
Graph View — Note Connections
An interactive visualisation of the links between notes in the vault. Nodes represent notes. Edges represent explicit [[wikilinks]] between them. Node size reflects link popularity. Clicking a node loads that note's detail. With the AI Thinker Vault pattern, the graph becomes a map of how your agents think — showing which concepts are most connected and where knowledge gaps exist.
Backlink Explorer
For any open note, the Backlinks tab shows all other notes in the vault that link to it. This is the reverse of the note's own forward links and is essential for understanding how a piece of knowledge is referenced across the vault. With agent-curated notes, backlinks reveal how the curation agent connected a new insight to existing knowledge.