Priority Triage
Use these tiers when promoting ActivityTracker / April (Active) material into Gemma Index notes. Mechanism and lived systems first. JD, OSINT, and relationship dumps last or never.
P1 — High (publish first)
Genuine density around:
- Mechanism — how something works (retrieval, search, learning loops, health/physiology mechanisms)
- Lived systems — how Nitish actually operated (personal knowledge capture, local chat search, workflows that were used)
- Learning / productivity / biohacking — only when there is real depth, not thin clippings
From this inventory (suggested):
| Item | Why |
|---|---|
nitish-capture/.../chat.txt (+ INDEX / MESSAGE_MAP / meta as one bundle) | Lived system: local AI chat search / personal knowledge retrieval |
| Dated MFine reports with unique dates (May 2026, June 2026, Nov 2025, 2023 set) | Lived health primary record; bio context — sensitive; index for personal use first, publish only after redaction decision |
P1 does not mean “publish raw medical PDFs publicly.” It means index first for personal knowledge value.
P2 — Medium
Useful supporting research and content frameworks.
From this inventory (suggested):
| Item | Why |
|---|---|
Research Strategy Development.pdf | Research / framing support |
ContextSwitchAI/...High-Ticket_Client_Outreach....md | Content / outreach framework chat (supporting, not core mechanism) |
P3 — Low / defer
Product specs as essays, raw JD material, OSINT, relationship dumps, thin admin paper, finance/HR paperwork not needed for mechanism writing.
From this inventory (suggested):
- Contact Details of HR & Manager.pdf
- Smartlead offer confirmation.pdf
- HR Sejal mail response (sensitive employment/health disclosure)
- Document-eSign Document.pdf / esign.pdf
- Income Shield - CRED.pdf
- Salary slip (also often Skip for publish)
Defer until P1/P2 Index notes exist. Do not batch these ahead of mechanism work.
Skip / never
Empty stubs, duplicates, vault junk, plugin shells, pure binaries.
From this inventory:
- All 0-byte daily / date stubs (
01 July.md,02 July.md,03 July.md,26 June.md,29 June.md,2026/20260521.md,8th May, Fri.md) attachments.txt/events.txt(2-byte empties)- Make.md / TaskNotes / Tags / Excalidraw
.spaceshells and.baseviews - Empty folders listed in Master Index (no content files)
- Syncthing installer tree,
.dmg, HTML plugin dumps, LinkedIn JSON/CSV (out of md/txt/pdf scope) - Likely duplicate MFine PDFs with identical/near-identical sizes (
mfine_doc*cluster) — keep one canonical after byte/hash check; mark rest[dup] - Salary slip and raw PII dumps for any public-facing publish path
Quick decision rules
- If empty or plugin shell → Skip.
- If mechanism / lived system / dense learning → P1.
- If supporting research or content framework → P2.
- If JD / HR / admin / OSINT / relationship → P3 or never.
- If duplicate of a better file → Skip after noting the keeper.
- Do not over-promote medical PDFs to “Authority publish” without evidence review; prefer Experiential / personal archive labeling until mechanisms are extracted.