Research
Zotero, Obsidian, and Local Hermes support paper intake, source notes, reading sessions, project intake, and idea capture.
- Project intake drafts
- Source notes
- Focused reading sessions
Hermes Research OS is a private workflow layer for literature review, bioinformatics learning, writing, academic administration, schedule control, and automation oversight. This public page shows only high-level status, project maps, and operating intent.
Zotero, Obsidian, and Local Hermes support paper intake, source notes, reading sessions, project intake, and idea capture.
A structured learning track connects wet-lab expertise with Python, R, statistics, NGS workflows, and reproducible analysis.
VPS Hermes, n8n, Google Calendar, and Google Tasks handle recurring briefs and review flows.
Local Hermes and VPS Hermes coordinate through a controlled handoff channel with thread-only replies and mirror guard rules.
Use when the answer belongs to the current Slack discussion and no background automation is needed.
Example: confirm receipt, answer one question, or ask for one missing detail.
Use when a local research result should become a Calendar, Task, Gmail, or n8n action.
Example: send title, date, deadline, owner, and expected action; do not send the whole source.
Use when the summary helps interpretation but should not be copied as the actual reply.
Example: mark Summary as context and ask for a status update, not a restatement.
The site structure borrows three public patterns: a portfolio guide index, a bioinformatics wiki term index, and a case-study review board for complex analysis workflows.
Instead of a marketing landing page, this page works like a compact operating manual.
A quick decision table replaces a long explanation.
| Need | Start here | Next surface |
|---|---|---|
| Read or summarize a paper | Zotero and Local Hermes | Obsidian Source Note |
| Turn an email into a task | Gmail and n8n filter | Slack approval, then Tasks |
| Plan the day | Calendar and Tasks | VPS Hermes daily brief |
| Study bioinformatics | Bioinformatics Session | Research Next Action and progress note |
A Blake-style guide index turns the system into routes. Each guide starts with one action, one source of truth, and one safety boundary.
Move from Zotero metadata and PDFs into one Obsidian Source Note, then decide whether the paper becomes a project, background source, or archive item.
start: paper intake
Start from one term, one data object, one validation risk, and one tiny example. Do not install a tool until its input, output, and report are understood.
start: term session
Draft in Markdown, polish in Notion when useful, and keep citation-critical material tied back to Zotero and source notes.
start: outline
Extract sender, deadline, amount, date, location, attachment, and required action before creating a task or calendar item.
start: structured fields
Let n8n handle repeated triggers, VPS Hermes handle scheduled checks, and Slack handle approval. Local Hermes should not run always-on automation.
start: approval path
Use the handoff channel when desktop context needs server automation. Reply in thread, avoid copied prompts, and log meaningful changes.
start: handoff id
An IncoDOM-style wiki surface keeps learning scannable: hot terms, recent study additions, and suggested next concepts stay separate from full notes.
A Vibe Genomics-style case board makes ambitious workflows reviewable without turning them into unsafe protocols or private-data demos.
The learning system now uses a term-first map: every session connects one core term to data objects, analysis workflow, math/statistics, wet-lab interpretation, and AI modeling.
Terms are ranked by how often they appear in real bioinformatics work and how much confusion they remove.
Each study note should be small enough to finish, but connected enough to become reusable research memory.
Vibe Genomics is treated as a workflow example, not as a protocol to reproduce. The useful structure is goal, raw data, secondary analysis, variant calling, validation, privacy, cost, and report.
Every bioinformatics result is read with its threshold and denominator. A score such as GQ, FDR, coverage, or accuracy is only useful when the selected subset and validation source are visible.
The learning map uses the wider IncoDOM index across NGS, RNA-seq, single-cell, microbiome, structure, statistics, and AI terms. Docking is only one example inside the structure track.
IncoDOM provides term signals, Reddit provides study-order signals, Biohackers provides frontier risk signals, and Vibe Genomics provides the report pattern. None of these are copied as-is.
The local curriculum keeps a generated inventory of IncoDOM terms and clusters them into P0/P1/P2/P3 study priority. A refresh tool updates the list before new study sessions are planned.
Popular repositories and recurring community advice are converted into a read-first tool map. High stars do not mean immediate installation; every tool is tied to file objects, reports, and validation risks first.
Learning materials are grouped into sequencing basics, qPCR/statistics, math for machine learning, structure and docking, imaging analysis, and LNP/AI modeling. Each session uses one short source segment, then verifies context before citation.
Public datasets, toy examples, and already-approved local materials come first. Personal genome interpretation, clinical claims, raw protocol execution, and private data exposure remain outside the public surface.
When a Hermes bot does not answer or local logging fails, a read-only readiness check distinguishes required handoff membership from optional direct-posting channels.
Study notes separate hard blockers from soft blockers. Missing reference data, wrong tool versions, low memory, unclear sample IDs, and retry notes become part of the learning artifact.
Each map shows the public-safe structure of a workflow: where information starts, which Hermes or n8n layer handles it, and where review or knowledge storage happens.
Recurring schedule and task review without exposing private calendar details.
Research papers move from collection into structured notes and review-ready summaries.
Email signals become concise review requests before any task is created.
Project ideas, experiments, writing, and admin work stay organized across tools.
Books and examples are converted into practical analysis skills and daily progress.
Public or toy analyses become stronger when each result includes its threshold, denominator, validation source, and retry notes.
Desktop research work can request always-on automation without sharing local internals or turning request text into a copied reply.
This page must not expose tokens, webhook secrets, private file paths, internal execution links, or personal schedule details.
Operational details live in private notes, internal chat logs, and the automation runtime. This page is only a public doorway.