- VAULT_MAP.md: comprehensive guide for LLMs and agents covering people, folder structure, naming conventions, frontmatter taxonomy (15 categories + 25 types), disambiguation rules, and search info - README.md: updated with pointer to VAULT_MAP.md
6.6 KiB
Vault Map
Personal knowledge base and document archive for Yanxin Lu and family.
If you are an LLM or agent working with this vault, read this file first.
People
| Name | Relation | Aliases |
|---|---|---|
| Yanxin Lu | Primary owner | lu |
| Xuewei Jiang | Spouse | jiang, Erica |
| Fengqin Xue | Mother-in-law | xue |
| Xianlu Jiang | Father-in-law | |
| Argo | Family cat |
Vault Structure
home.md — Personal dashboard (Obsidian Tasks widget)
VAULT_MAP.md — This file. Vault guide for humans and LLMs.
README.md — Git repo description.
copilot/ — Obsidian Copilot plugin data. Do not modify.
Readwise/ — Readwise sync plugin data. Do not modify.
templates/ — Note templates for journals, book notes, etc.
journals/ — Daily journal entries (2022-present)
notes/ — Personal notes: books, martial arts, tech setup
trip_plans/ — Trip itineraries and travel planning
writing/ — Long-form writing projects
documents/ — Document archive (see below)
Document Archive (documents/)
The documents/ folder is an archive of personal documents (PDFs, images, etc.) with markdown transcriptions alongside each file. The entire vault (not just documents/) is indexed by the notesearch skill (LlamaIndex + Ollama vector search).
Folder Map
| Folder | Contents |
|---|---|
academic/ |
PhD thesis, conference presentations, coursework, papers |
argo/ |
Vet records for family dog |
car/ |
Vehicle purchase documents |
dmv/ |
DMV applications and supporting docs |
employment/ |
Job offers, onboarding docs (subfolders by employer) |
finance/ |
Brokerage accounts, banking |
house/ |
Home purchase (grant deed, inspection, renovation, quotes) |
id/ |
Identity documents: DLs, passports, SSNs, China IDs, marriage certs (subfolders by person) |
immigration/ |
H-1B petitions, green card, visa stamping, travel visas, passport renewal |
insurance/ |
Auto and medical insurance |
lease/ |
Rental leases, addenda, vacate notices |
medical/ |
Medical records, lab results, prescriptions (subfolders by provider or condition) |
receipts/ |
Purchase receipts, service receipts |
tax/ |
Tax returns, W-2s, 1099s, 1098s, property tax (subfolders by year) |
travel/ |
Flight itineraries, customs forms, EVUS, family visit docs |
wedding/ |
Wedding-related documents |
Transcription Convention
- Each original file has a matching
.mdtranscription alongside it (e.g.,lu_w2_meta_2025.pdf+lu_w2_meta_2025.md) - The
.mdfile is a faithful full transcription of the source — every data point preserved, not a summary - Exception:
immigration/subfolders use a singlesummary.mdper subfolder instead of per-file transcriptions, because immigration petitions reuse many of the same supporting docs
File Naming Convention
{person}_{doctype}_{year}.{ext}
- person:
lu,jiang,nichol,xue, or omit if joint/obvious - doctype: descriptive snake_case (e.g.,
w2_meta,dl_ca,eye_exam) - year: 4-digit year, or
YYYY_MMif needed for disambiguation - ext: always lowercase
Frontmatter Taxonomy
Every .md transcription has YAML frontmatter with these fields:
---
category: tax
type: tax-form
person: Yanxin Lu
date: 2025
source: lu_w2_meta_2025.pdf
---
Categories
Category always matches the top-level folder under documents/.
| Category | Folder |
|---|---|
academic |
academic/ |
car |
car/ |
dmv |
dmv/ |
employment |
employment/ |
finance |
finance/ |
house |
house/ |
id |
id/ |
immigration |
immigration/ |
insurance |
insurance/ |
lease |
lease/ |
medical |
medical/ |
receipts |
receipts/ |
tax |
tax/ |
travel |
travel/ |
wedding |
wedding/ |
Types
| Type | Used for |
|---|---|
id |
Driver's licenses, passports, SSNs, China IDs, visa stamps, EADs |
marriage-cert |
Marriage certificates and translations |
tax-return |
Federal and state tax returns (1040, 540, etc.) |
tax-form |
W-2, 1099, 1098, 1098-T, 1042, 1095-C, 8833, extensions, IRS transcripts |
property-tax |
Property tax receipts |
registration |
Vehicle registration |
lease |
Rental leases, addenda, vacate notices, credit checks |
deed |
Grant deeds, title reports |
contract |
Service contracts (remodel, DJ, etc.) |
inspection |
Home inspections, PID/NHD reports |
financial |
Bank statements, loan docs, preapprovals, brokerage applications |
offer-letter |
Job offers |
lab-results |
Blood work, semen analysis, exams, HSG, genetic screening |
prescription |
Eye prescriptions, medications |
vaccine |
Vaccine cards, declarations, COVID test results |
medical-form |
Intake forms, oral surgery forms |
immigration |
H-1B, I-140, I-485, PERM, passport renewal (in summary.md) |
travel-visa |
Visitor/Schengen visa applications (in summary.md) |
visa-stamping |
Consular H-1B visa stamping (in summary.md) |
academic |
Thesis, papers, presentations, posters, coursework |
receipt |
Payment receipts, invoices, shipping |
quote |
Price quotes |
blueprint |
Architectural drawings |
travel-doc |
Flight itineraries, customs forms, EVUS, invitation letters, Global Entry |
vet |
Vet medical records |
Disambiguation Rules
When a document could belong in multiple folders, file by primary purpose:
| Document | Goes in | Not in |
|---|---|---|
| W-2, 1099, 1098 | tax/{year}/ |
employment/ |
| Vehicle registration (for tax) | tax/{year}/ |
dmv/ |
| Passport/visa copy (for immigration filing) | immigration/ (as part of petition) |
id/ |
| Passport/visa copy (standalone) | id/{person}/ |
|
| Vaccine card (standalone) | medical/covid/ |
|
| Vaccine card (for travel) | travel/ |
|
| Insurance card | insurance/medical/ |
medical/ |
| Supporting docs gathered for a specific application (DMV, immigration) | Stay with that application | Don't split out |
Search
The vault is indexed by notesearch (LlamaIndex + Ollama embeddings). Search is semantic — query by meaning, not keywords. The index lives at .index/ and is rebuilt on the blade server after git pull.
Only .md files are indexed. Original binary files (PDFs, images) are not directly searchable but their content is captured in the .md transcriptions.