Orcanium Agent

Orcanium

**Multi-agent runtime** — run multiple agents in one process, each with its own LLM provider, personality, memory, knowledge base, schedule, and messaging channel.

Orcanium is the open-source, locally runnable agent runtime: CLI, terminal UI, REST API with SSE streaming, messaging channels, plugin system, and local web dashboard. It is independent from Orcanium's private services — local and BYOK operation require no account.

Run Multiple Agents, Without the Complexity

---

Every agent is a named entity with fully isolated state — config, provider, personality, memory, and channel. One process, many distinct assistants, zero shared-state surprises. Conversation flows through an intent-classification → retrieval → attention-scoring → prompt-assembly pipeline, then an LLM → tools → LLM loop backed by 95+ built-in tools and an event bus.

Bring your own keys, run fully local, or mix both. No account, no vendor lock-in.

---

🚀 Quick Start

One-line install — downloads the latest release tarball to `~/.orcanium/` (no git needed):

curl -fsSL https://orcanium.com/install.sh | bash

Then start your first agent session:

orcanium setup
orcanium run

---

⚡ Core Features

Feature	Capabilities
Multi-agent	Create and run multiple agents with per-agent LLM config, personality, memory, knowledge, schedule, and messaging channels — all in one process.
Multi-provider LLM	OpenAI, Anthropic, Gemini, Ollama, OpenRouter, Groq, Together, Fireworks, DeepSeek, LM Studio.