Any model key
Add provider keys once. Start sessions against the model that fits the task instead of changing apps.
Chat with OpenAI, Claude, DeepSeek, OpenRouter, Cerebras, and local models from one TUI. Skills, memory, tools, cron, and gateways included.
Arcana does not sell a model. It gives your model keys a terminal-native session layer with memory, tools, and scheduled runs.
OpenRouter sells unified model access. Arcana sits one layer above: the terminal surface where keys, tools, memory, jobs, and gateways meet.
Add provider keys once. Start sessions against the model that fits the task instead of changing apps.
Search previous sessions, facts, artifacts, and decisions. Keep context outside one provider's chat history.
Run filesystem, shell, git, gateway, and skill tools from explicit profiles. The tool surface is visible.
Load specialized instructions only when needed. Keep the base session clean.
Schedule recurring prompts, reports, reviews, and checks from the same terminal runtime.
Expose sessions through Telegram, Discord, Slack, and WhatsApp when the terminal should not be open.
No account required for local use. The first run should feel like using a developer tool, not signing up for another workspace.
Run Arcana from npm. Keep it local.
npx arcana-ai@latest
Connect the provider you already pay for.
arcana keys add openrouter
Chat, run tools, save memory, or schedule a job.
arcana
Model costs stay with your provider. Arcana charges for the terminal interface, session layer, sync, relay, and team surfaces.