# ARCANA

ARCANA is a terminal-native AI engineering workbench built for developers who want to use AI without giving up inspection, control, or evidence. It combines a CLI/TUI workflow with model choice, scoped tools, persistent memory, routing, cost visibility, and verifiable execution. You bring your own supported provider keys when you want direct provider billing; Arcana can also provide optional proxy infrastructure where documented.

## What ARCANA is for

ARCANA is designed for engineering sessions in which the operator needs to understand what the model did, which tools it used, what changed, what was allowed or denied, and what evidence remains afterward. Instead of treating an agent run as an opaque conversation, Arcana treats the session as inspectable engineering work.

### Model choice and routing

Use supported model providers from one terminal workflow. Routing surfaces can expose model selection, cache behavior, token use, cost, and related telemetry so model choice can be evaluated rather than guessed.

### Scoped execution

Tool access is intended to be explicit and bounded. ARCANA surfaces permissions and approvals around actions that can affect files, commands, external services, or other execution boundaries.

### Memory and continuity

ARCANA provides local memory and storage in the Community edition, with additional memory and synchronization capabilities available on paid infrastructure tiers. The goal is to preserve useful engineering context without forcing every session to start from zero.

### Evidence and inspection

ARCANA is built around the idea that AI-assisted engineering should leave inspectable evidence. Session history, execution details, routing information, and related records help developers review what happened after the model response is gone.

## Start here

- Install ARCANA: https://arcana.otnelhq.com/install
- Read the documentation: https://arcana.otnelhq.com/docs
- Explore models: https://arcana.otnelhq.com/models
- Review routing: https://arcana.otnelhq.com/routing
- Compare plans: https://arcana.otnelhq.com/pricing
- Check service status: https://arcana.otnelhq.com/status
- Read the machine-readable API: https://arcana.otnelhq.com/openapi.json

## For agents and automated clients

Use https://arcana.otnelhq.com/llms.txt for concise retrieval guidance. The canonical homepage supports `Accept: text/markdown`, and the REST surface is described by OpenAPI. Unsupported paths return real HTTP 404 responses rather than the homepage shell, so clients can distinguish missing resources from valid ones.