How to use Hermes Agent to compile an Obsidian wiki
Jack Ivers runs Hermes on a Mac Mini to turn bookmarks, media and meeting notes into linked Obsidian pages, reviewing health reports and skill changes.
cto4.ai
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Compile [captured sources] into a linked Obsidian knowledge base using [ingestion boundaries] and separate [namespaces]. Track each source through a work queue, preserve provenance and report failed jobs. Include agent-edited skills in a daily diff report so I can review changes before deployment.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Jack Ivers
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Author reports running knowledge refinery processing sources into linked pages and client namespaces.
- Human involvement
- Author specified architecture/tests, reviewed overnight skill diffs over coffee and committed changes.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Ingested sources into Kanban cards via simple scripts.
- Dispatcher/parallel workers compiled entity/concept pages directly into vault.
- Cron reports exposed health and skill edits.
Jack Ivers turned an idle Mac Mini into a Hermes Agent knowledge refinery for his Obsidian vault. Incoming material included bookmarks, media and meeting transcripts. The aim was to turn scattered captured sources into linked entity and concept pages that could be maintained as new information arrived.
Simple ingestion scripts created source cards rather than spending model calls deciding what should enter the queue. A dispatcher and parallel workers compiled those sources into the vault. The same machinery supported separate namespaces for his AI-industry knowledge, consulting clients and personal material. Meeting transcripts arrived through a connector he controlled.
Running continuously introduced operational work. Daily Discord check-ins reported health and identified skills the agent had edited in place. Ivers describes reviewing a repaired bookmark-sync import and new skill notes over coffee, then committing those changes. He also maintained authentication and dealt with provider outages; a Git push alone did not prove the running machine had deployed safely.
The source reports a functioning personal system, while company-wide access control remained unfinished.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Blog · Jack Ivers
Frequently asked questions
How does new material enter the workflow?
Simple ingestion scripts create source cards; a dispatcher and parallel workers compile them into linked knowledge pages.
How are agent-authored skill changes reviewed?
Daily Discord reports identify in-place skill edits. Ivers describes reading the diffs and committing accepted changes.
Was it ready for a whole company?
No. The author says safe company-wide access control remained unfinished, even though separate client and personal namespaces already ran.
More details
Remaining steps
- Built namespace isolation for client knowledge bases.
About this account
Hermes workers compiled queued sources into linked vault pages, operated scheduled reporting and edited skills; the author reviewed changes and maintained the system.
Primary first-person account opened and checked on 2026-10-14. Maintains OAuth and provider failures; unattended writes and skill drift require tending. Team access-control version not yet achieved. Outcomes and test claims are attributed to the source and were not independently reproduced. Prompt is suggested by Agents In Use, not the author's quoted prompt.
Credit: Jack Ivers on Blog
Source published
Added to Agents In Use
Source checked
Tools and topics
Hermes Agent · Obsidian · Mac Mini · Kanban · Discord · Git
Knowledge compilation · Obsidian · Operational review
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.