LobeHub - Your Chief Agent Operator for multi-agent work
by•
LobeHub is a Chief Agent Operator (CAO) that builds, runs, and coordinates your AI agent team. Describe a goal, and it assembles the right agents/skills, runs tasks in parallel in the cloud, routes work across models, and reports back only when decisions are needed—via your existing channels (Slack/Discord/Telegram/iMessage). Less tab-switching, more outcomes.
Replies
Is iMessage support live? That’s the first time I have seen a launch support Apple’s native chat.
LobeHub
@antonio_manuel1 will support it in the few weeks. 🔥 Stay tuned!
LobeHub
😁 273K Skills + 51K MCP servers behind one prompt feels a little unreal even to me. Let me know what you end up running through it — I want to see the weird stuff.
LobeHub
@cy948 Haha for real, the numbers sound fake until you actually wire something up and watch it route 😄
BetterClaw
273K skills sounds impressive but how many of those have actually been vetted for security. running untested skills across parallel agents is a huge attack surface especially when you're routing through real credentials on slack and gmailv
LobeHub
@tina_chhabra We're building a security audit for Skills in marketplace, because making sure the entire ecosystem is secure enough is one of our commitments!
SpotVault
Have been using it for icons. Really cool.
LobeHub
@samihindi Thanks! Have fun with icons~ 😆
With 273K+ skills in the pool, how does CAO decide which agents to actually assemble for a goal — is there a ranking or filtering layer, or does it try a broader set and prune based on early results?
LobeHub
@hirogure Great question — we don't load all 273K into context. The agent only sees a lightweight catalog (identifier + one-line description) of what's installed, picks what fits via an Activator call, and only hits the marketplace's ranked searchSkill API when nothing local matches. So it's rank-and-filter upfront, not broad fan-out then prune — picking wrong is much cheaper than executing wrong.
The CAO framing clicked — im tired of being the human router between claude code and slack pings
LobeHub
@novamaker01 Haha "human router" is painfully accurate — that was literally the whiteboard sketch that started this whole thing. Glad the framing landed 🫡
wait so its like... i just tell it what i want to do and it goes off and does it? not sure i'm brave enough to let an AI just run 5 agents at once in parallel without watching what's happening lol
LobeHub
@boyuan_deng1 It's designed to do task like this. Just give a try.
Multi-model routing sounds like a huge advantage. Are users able to choose preferred AI models for certain workflows?
LobeHub
@elijah_smith6 yes , Absolutely
LobeHub
@elijah_smith6 You can pin a preferred model per agent, per skill, or per workflow step — e.g. Claude for long-context reasoning, GPT for structured output, a cheap fast model for classification/routing. CAO respects those pins by default.
When nothing's pinned, it auto-routes based on the task profile (context length, latency budget, cost ceiling, tool-use needs). And you can always override mid-run if you want to swap models for a specific step.
The agent coordination layer feels futuristic. Can agents collaborate with each other dynamically during long tasks?
LobeHub
@ethan_thompson3 Yeah, that's basically the whole point of CAO 👇
Agents pass context and intermediate outputs to each other mid-task, and CAO can spin up new agents on the fly if a step needs a capability the current crew doesn't have. You stay in the loop only when a real decision needs you — everything else just flows.
The "Chief Agent Operator" concept resonates. I run 15+ automated agents (uptime monitoring, social media engagement, security audits, competitor analysis) and the coordination layer is what took the longest to build. Getting agents to read each other's outputs and prioritize actions without conflicting recommendations was months of iteration.
The daily briefing approach is smart — my system does something similar with a "Manager" agent that aggregates all overnight findings into one executive summary. How does LobeHub handle conflicting recommendations from different agents?
LobeHub
@ytubviral Respect — 15+ agents in production is no joke, you've clearly done the hard miles 🫡
On conflicts: CAO doesn't try to auto-resolve them. When two agents disagree on what to do next, it pauses, surfaces both recommendations with their reasoning, and asks you to call it. The bet is that conflicts usually mean the goal itself needs clarifying — not that one agent is "wrong." Silent auto-merging is where trust dies.
Sounds like your Manager agent setup is doing similar work — would genuinely love to compare notes sometime.