Any model key
Add provider keys once and switch models by task, not by app.
Use OpenAI, Claude, OpenRouter, DeepSeek, Cerebras, and local models from one terminal. Arcana adds memory, tools, and scheduled runs on top of your existing keys — no new provider, no vendor lock-in.
Arcana does not sell a model. It gives your existing provider keys a terminal-native session layer with memory, tools, and scheduled runs.
OpenRouter unifies model access. Arcana sits one layer above — the terminal where keys, tools, memory, jobs, and gateways work together.
Add provider keys once and switch models by task, not by app.
FTS5-powered session memory with extracted facts, full-text search, and optional cloud sync across machines.
Run shell, filesystem, git, gateway, and skill tools through explicit profiles with visible permissions.
Load specialized instructions only when needed to keep the base session clean.
Schedule recurring prompts, reports, reviews, and checks from the same runtime.
Route sessions to Telegram, Discord, Slack, or WhatsApp when the terminal is not the right surface.
7 arcane themes with custom background images for supported truecolor terminals.
Push local facts to the cloud, pull on other machines, and share knowledge across your workspace.
No account is required for local use. Arcana should feel like a developer tool from the first run.
Run Arcana locally from npm.
npx arcana-ai@latestAdd your Pro license when you need advanced features.
arcana license activate <key>Chat, run tools, save memory, or schedule a job.
arcanaModel costs stay with your provider. Arcana charges for the terminal layer, memory services, sync, relay, and team surfaces.
arcana credits buy.