Vault Schema

Vault Schema

Alfred organizes vault records into 20 entity types plus 5 learning types. All records are Markdown files with YAML frontmatter and are linked together with Obsidian [[wikilinks]].

Record Types

Operational Records

Type	Directory	Purpose	Key Fields
project	project/	Bounded initiative with a goal	status, client, owner, parent, location
task	task/	Action item or to-do	status, project, assignee, due_date
session	session/	Work session log	status, project, participants
conversation	conversation/	Multi-turn exchange (meeting, chat, email)	status, participants, project
input	input/	External input (email, document, form)	status, source, project
note	note/	Free-form note or analysis	status, project, subtype
process	process/	Repeatable workflow or procedure	status, owner
run	run/	Instance of a process execution	status, process, project
event	event/	Scheduled or past event	status, date, location, participants
thread	thread/	Discussion thread	status, project

Entity Records

Type	Directory	Purpose	Key Fields
person	person/	Individual person	status, org, role, email
org	org/	Organization or company	status, org_type
location	location/	Physical place	status, address
account	account/	Service account or subscription	status, provider, org
asset	asset/	Physical or digital asset	status, owner, location

Learning Records (Epistemic)

Type	Directory	Purpose	Key Fields
assumption	assumption/	Belief being operated on	status, confidence, claim, evidence_excerpt
decision	decision/	Choice made with rationale	status, confidence, claim, evidence_excerpt
constraint	constraint/	Hard limit identified	status, confidence, claim, evidence_excerpt
contradiction	contradiction/	Conflicting information	status, confidence, claim, evidence_excerpt
synthesis	synthesis/	Pattern connecting multiple observations	status, confidence, claim, evidence_excerpt

Status Values

Each record type has a defined set of valid statuses:

Types	Valid Statuses
project	active, paused, completed, abandoned, proposed
task	todo, in-progress, done, blocked, cancelled
session, conversation, run	active, completed
person, org, location, account, asset	active, inactive
assumption, constraint	active, retired, superseded
decision	final, draft, superseded, reversed
contradiction	unresolved, resolved
synthesis	draft, final

Required Fields

Every record must have:

• type — the record type (one of the 20 types above)
• name (or type-specific name field like subject for decisions) — the record title
• status — current status
• created — ISO date of creation

Relationship Fields

Records connect through multiple relationship types:

Field	Type	Purpose
related	list	General relationships ([[type/Name]])
relationships	list	Typed relationships (added by surveyor)
supports	list	What this record enables
based_on	list	Assumptions/decisions this rests on
depends_on	list	Operational prerequisites
blocked_by	list	Active blockers
approved_by	list	Authority chain (person links)

All relationship values use wikilink syntax: [[type/Record Name]]

Templates

Each record type has a template in the vault's _templates/ directory. Templates define:

1. Default frontmatter — all fields with sensible defaults
2. Body structure — heading, description placeholder
3. Base-view embeds — Dataview sections like ![[project.base#Tasks]]

When a record is created (via alfred vault create or the curator pipeline), the template is automatically applied. Even when custom body content is provided, base-view embeds are preserved and appended.

Template Placeholders

• {{title}} — replaced with the record name
• {{date}} — replaced with today's ISO date

Base Views

Base views live in _bases/ and contain Dataview query definitions. They are embedded into records via Obsidian's transclusion syntax:

## Tasks
![[project.base#Tasks]]

This renders a live table of all tasks linked to the current project. Each record type has its own base file with relevant sections.

Example: project.base

Sections in project.base:

• Assumptions, Decisions, Constraints, Contradictions — learning records linked to this project
• Dependencies — projects this depends on
• Tasks — tasks assigned to this project
• Sub-projects — child projects
• Sessions, Conversations, Inputs, Notes — operational records linked to this project
• Learnings — all learning records referencing this project

User Profile

The user-profile.md file in the vault root helps Alfred understand who the vault owner is. This enables context-aware entity extraction — the curator will only create records for entities relevant to the vault owner, not every person, company, or topic mentioned in source material.

The file is created during quickstart with a template. Fill it in with your name, work context, interests, and optionally things to ignore.