Vault Operating Rules

Purpose

This vault is the retrieval-facing knowledge graph for Vitals.

It is not the raw research archive and not the master planning layer.

The 3-layer model

1. Source layer

Location:

• research/

Contains:

• worker JSON packets
• briefs
• crawls
• partial outputs
• CSV seed corpora

Use for:

• provenance
• evidence tracebacks
• re-synthesis
• auditing claims

Do not optimize this layer for Obsidian readability.

2. Deep research library

Location:

• skills/knowledge-base/

Contains:

• long-form compound docs
• protocol docs
• research plans / indexes

Use for:

• canonical long-form synthesis
• dense background reading
• detailed mechanism and protocol context

This layer is source-adjacent, but more human-readable than raw packets.

3. Retrieval graph

Location:

• obsidian/Vitals-KB/

Contains:

• hub notes
• MOCs
• mechanism notes
• biometrics notes
• recovery/risk notes
• detection-model notes
• interaction notes

Use for:

• LLM retrieval
• fast human lookup
• cross-note synthesis
• Obsidian graph traversal

What belongs in the vault

Yes

• high-value substance monographs
• high-value compound / peptide monographs
• reusable mechanism notes
• reusable biometric signatures
• recovery / risk notes
• detection-model notes
• meaningful stack / interaction notes
• MOCs and knowledge maps

No / usually no

• raw worker JSON
• crawl artifacts
• CSV corpora
• giant planning docs
• prompt packs
• duplicate legacy formats
• every tiny subtopic as its own note

Note types

Hub notes

Folder:

• 01-Substances/ for substances
• 06-Peptides-and-Interactions/ for peptide/compound hubs (current convention)

Hub note should include:

• YAML frontmatter
• TL;DR
• key facts
• why it matters for Vitals
• summary of mechanisms
• summary of biometrics / recovery / risks
• links to child notes

MOCs

Folder:

• 00-Maps/

Purpose:

• pure navigation
• minimal prose
• no duplication of the hub note

Mechanism notes

Folder:

• 02-Mechanisms/

Create only when the mechanism is reusable across multiple notes or retrieval queries.

Biometrics notes

Folder:

• 03-Biometrics/

Use for:

• HRV
• RHR
• sleep architecture
• REM suppression
• cardiovascular signatures
• recovery distortion

Recovery / risk notes

Folder:

• 04-Protocols-and-Recovery/

Use for:

• withdrawal
• crash states
• hangover mechanisms
• psychosis risk
• CHS
• other meaningful consequence notes

Detection-model notes

Folder:

• 05-Detection-Models/

Use for:

• wearable detection logic
• confounders
• feature stacks
• realistic detection limitations

Interaction notes

Folder:

• 06-Peptides-and-Interactions/

Only create interaction notes when the interaction is meaningful enough to retrieve on its own.

Naming rules

Prefer these patterns

• Cannabis.md
• Alcohol.md
• Cocaine.md
• Cannabis MOC.md
• CB1 receptor.md
• REM suppression.md
• Cannabis detection model.md
• Cannabis peptide interactions.md

Avoid

• duplicate names for the same concept
• multiple phrasings for one note
• inconsistent connector styles (and, &, vs) unless meaningfully different

Canonical patterns

• MOCs: [Topic] MOC.md
• detection notes: [Topic] detection model.md
• interaction notes: [Topic] peptide interactions.md or [Topic] interactions.md
• mechanism notes: concept-first names, not essay titles

Conversion rules

When converting long-form research into vault notes:

1. keep the graph small but useful
2. split by reusable concept, not by arbitrary paragraph count
3. prefer wikilinks everywhere
4. preserve high-signal numbers and claims
5. do not bring over every citation into every note
6. do not create notes for trivial or one-off concepts
7. if a concept is only useful inside one hub, keep it inside the hub

Duplicate handling

If duplicate notes exist:

1. choose one canonical note title
2. migrate stronger content into the canonical note
3. update backlinks
4. remove or archive the duplicate

Default workflow for new topics

1. research packet / source material created in research/
2. long-form synthesis exists or is generated if needed
3. spawn MiniMax for Obsidian graph build
4. run vault QA for duplicate names / broken links / obvious overlap
5. add note into MOCs / maps

Retrieval principle

The vault should optimize for:

• short retrieval paths
• semantic clarity
• concept reuse
• cross-topic linking

It should not try to replace the source layer.