Course promo
Hermes Command Desk: One Dashboard for Chat, Voice, Tasks and Memory
A screen tour of a custom Hermes dashboard the creator calls a "command desk", which puts chat, a voice agent, a Kanban-style task board, scheduled jobs, skill management and one-click model switching behind a single UI, with an Obsidian vault as the shared memory layer every agent reads from. The video is a feature walkthrough rather than a build guide — no installation commands or configuration are shown, and it resolves into an offer to download the assembled system from the creator's paid community.
Concepts covered, if you want to build the equivalent yourself: a shared memory store so switching models doesn't mean re-explaining context; a task queue that dispatches work to agents asynchronously; and a "goal mode" long-running loop. Our own coverage of those pieces is in Hermes memory, Hermes tasks, and the Hermes web dashboard.
Watch on YouTube → · ← Back to News
What you can actually set up from this
Extracted from the video's own transcript — the specifics the original summary left out.
Reproducible steps
- Change the main model from the dashboard
In his Manage tab you open models and change the main model in two clicks. He contrasts it with the terminal route, where you run Hermes's model selector and scroll a long list. He used this to add Muse Spark and Qwen 3.8 within days of release.
- File tickets instead of prompts
A kanban board takes tickets; a dispatcher spawns workers that build while you do something else. Tasks are triaged automatically and you monitor them from the dashboard.
- Use goal mode for long jobs
Type one goal and the agent loops for hours without you; you come back to the finished work.
- Share one memory across agents
Each exchange updates an Obsidian vault used as the shared memory, so Claude Code, Muse Spark, Qwen 3.8 and a free Claude Code setup can all pull from the same notes and you don't re-explain your context.
- Schedule recurring research
His Hermes Oracle job is scheduled on a cron job and was last updated five hours before the recording; the Manage tab also lists scheduled tasks, lets you add a skill or schedule a new job, and shows past sessions across platforms.
- Use free local models to avoid token cost
He says free local models such as LFM work in the system and that LFM was trained on Hermes Agent. You can hide dashboard modules you don't use.
Gotchas
- The speaker says Hermes has guardrails: for deletions or anything it is unsure about, it can ask you and come back to it later.
- The dashboard is the speaker's own setup, installed from a zip file sold through his community; none of it is a built-in Hermes feature.
- Model names and releases mentioned (Muse Spark 1.2, LFM 2.5, Qwen 3.8) are as he stated them.





