Appearance
Private research & notes
Problem: I have PDFs, markdown exports, and local notes I want synthesized without uploading to any cloud service.
Setup: Fully offline. Web tools disabled.
Use case 1: Summarize local documents
text
Read all .md files in docs/getting-started/ and write a single
"Getting Started Summary" page at docs/introduction/quick-summary.md.
One paragraph per page, plain language, no em dashes.What Hermes does: Reads each file, synthesizes, writes new file.
Why offline matters: These docs live on my machine. Nothing goes to an API.
Use case 2: Compare local notes
text
I have two approach notes:
- ~/notes/hermes-setup-v1.md
- ~/notes/hermes-setup-v2.md
Compare them. What's different? What's better in v2?
Output as a markdown table: Topic | v1 | v2 | Recommendation.Use case 3: Build a FAQ from session history
text
Search my recent Hermes sessions for anything related to "context length"
or "Ollama". Compile the top 5 questions and answers into
docs/best-practices/ollama-faq.md.What Hermes does: Searches ~/.hermes/sessions/, extracts patterns, writes FAQ.
Note: Quality depends on how much session history exists.
Use case 4: Index a folder of PDFs (with text extraction)
text
In ~/Documents/research/, list all PDFs. For each one, extract the
title and first 200 words. Create an index file INDEX.md with
filename, title, and summary.Requires PDF text extraction tools on your system. Hermes orchestrates via shell.
Tips
- Local models summarize well but may miss nuance on long docs. Split large files into sections.
- Ask Hermes to quote source lines when accuracy matters.
- Store research output in the project so context files pick it up next session.
Real outcome
This handbook's offline setup page started as notes Hermes synthesized from my Ollama config experiments and the official docs, all processed locally.