I am Octo.

AI agent. Real work. Your call.

$ curl -fsSL https://iamocto.com/install.sh | sh coming soon

macOS & Linux · MIT licensed · or build from source with cargo

42built-in tools

6chat channels

anyOpenAI-compatible model

1binary, in Rust

MITlicensed

Capabilities

I have hands, memory — and limits.

🧰

42 built-in tools

Shell, files, git/GitHub/Docker/SQL, web search, HTTP, email, payments, monitoring, screen capture. Full list →

🔌

Extensible by spec

Point it at an MCP server or an OpenAPI spec and the endpoints become tools. Teach procedures with one markdown skill file.

💬

Every channel you use

CLI, built-in web chat with streaming, Telegram, WhatsApp, Discord, Slack — a token in one config file each.

🧠

Any model, with a net

Ollama local, OpenAI, seeme.ai, vLLM, llama.cpp. Failover chains: primary errors, the next backend answers.

🤖

Awake when you aren't

Multi-agent swarms, missions, cron schedules, heartbeat self-monitoring, persistent memory that survives restarts.

🔒

Limits you set

Permission tiers, daily token & cost budgets, injection and leak scanning, append-only audit log — and octo kill.

Everything on this page ships today. Features outside this set are labeled preview in the docs — not advertised here.

Channels

I'm where you already are.

⌨️ CLI🌐 Web chat✈️ Telegram 💚 WhatsApp🎮 Discord🏢 Slack

Get started

Install me, pick my brain, put me to work.

Soon. The binaries are built and waiting on the release server — the door just isn't open yet. Memorize the one-liner; you'll want it.

curl -fsSL https://iamocto.com/install.sh | sh
octo setup        # Ollama / OpenAI / seeme.ai / custom — it lists your models
octo run "summarize this repository"

# when you want it always-on:
octo gateway      # daemon: web chat, channels, cron
octo doctor       # health-check the install
octo update       # checksum-verified updates, delivered by verr