Why now
Why AgentMux,
why now.
Agents already write production code and reshape live systems. Our tools still treat them like terminals.
01
The agentic era
outran its tools.
Agents read, write, test and iterate on their own. The workspace to drive them didn't keep up.
02
Subagents
run blind.
Swarm lists every subagent under its parent, with a live feed of what each is doing.
03
Your tools shouldn't
eat your RAM.
Chromium over a pure Rust backend: no Node runtime, no garbage collector pauses.
3–40
GB RAM reported · Cursor & Windsurf, long sessions04
Agents regress
silently.
Every tool call streams as it lands, so you catch the turn when it happens, not after it merges.
05
One workspace
for every provider.
Claude, Codex, Gemini and more as native panes, each with its own account and reusable Bundles.
06
Agents need to
talk to each other.
One agent sends a message, another reacts. Every message arrives tagged with its sender and trust level.
07
Agents should drive
their own UI.
Open panes, rename tabs and show custom views, all over a typed local API.
08
Oversight is
the requirement.
53% of organizations report AI agents exceeding their intended permissions (CSA/Zenity, 2026). The EU AI Act's high-risk obligations phase in from December 2027; the Colorado AI Act takes effect January 1, 2027.
09
AgentMux builds
AgentMux.
Agents running inside AgentMux wrote it: frontend, backend and infrastructure, in parallel.
Free · Open source · Apache 2.0
Try AgentMux.
Alpha software. Features may be incomplete or unstable. AI agents generate content that may be inaccurate — always review outputs. Report issues