{{ msg.who }}

{{ msg.text }}

{{ msg.routingTitle }}

{{ rt.model }} {{ rt.part }} {{ rt.prov }}

muse is thinking…

Fusion {{ fusionModeLabel }} {{ fusionModeHint }}

{{ m.label }}

{{ m.value }} {{ m.delta }}

{{ agent.name }}

{{ agent.role }}

{{ agent.statusLabel }}

Now {{ agent.nowText }}

{{ agent.tasks }} tasks handled today

Running now

{{ sk.name }}

{{ sk.desc }}

On

Skill library

{{ cat.label }} {{ cat.count }}

Put any models together into one blend, drawn from every provider you have connected. muse can add or drop models on its own as a task needs.

One omni source for every model. Connect any provider, add any model, and choose which ones muse can use — you are never locked to one.

{{ p.name }}OMNIFREE

{{ p.kindLabel }}

{{ g.label }}

{{ g.desc }}

{{ mt.label }}{{ mt.pct }}%

{{ mt.detail }}

muse switches how it works to fit the moment. Pick the one that should lead right now — it can still move between them on its own.

{{ m.name }} {{ m.badge }}

{{ m.desc }}

A deep bench of specialist roles muse convenes for audits, launch reviews, and big decisions — each department weighs in from its own angle.

{{ c.roles }}

{{ c.dept }}

{{ c.note }}

{{ t.name }}{{ t.countLabel }}

{{ t.desc }}

Same muse, every surface — it remembers the conversation as you move between them. Voice memos are transcribed automatically.

muse can run on your laptop, a server you own, or serverless cloud that sleeps when idle. Pick where the work should happen.

{{ r.name }}{{ r.tag }}

{{ r.note }}

Everything muse, in one place. Desktop, Android, and source are now published to a single rolling release that auto-refreshes from the latest main — the old per-platform channels are gone, but all of their downloads live here.

Every tagged version this build descends from. These are period-accurate records — they predate the muse rebrand and keep their original “Hermes Agent” naming.

{{ h.ver }} {{ h.date }}

{{ h.name }} {{ h.summary }}