> ## Documentation Index > Fetch the complete documentation index at: https://atcyrus.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # What Cyrus Manages Cyrus intentionally limits user-facing configuration. This page documents what is **automatically handled**, so you don’t have to think about it. ## Infrastructure **Self-host Runtime** * Local repository paths * Worktree lifecycle * Process management * Cloudflare tunnel connectivity **Cloud Runtime** * Hosted compute (dedicated VPS) * Networking and domains * Scaling and resource allocation All infrastructure is provisioned and maintained by Cyrus, regardless of your pricing plan. ## AI model selection Cyrus runs on high-quality defaults so you don't have to think about model selection. Each provider ships with a strong primary model and an automatic fallback, and most users will never need to change this. For the current list of supported providers, models, and override aliases (including `fable`, `opus`, `sonnet`, `haiku`), see [Default Model Provider](/providers/default-model-provider#supported-models-by-provider). ## Tool permissions Cyrus enables a safe, limited toolset for AI reasoning: * File read/write * Git operations * GitHub CLI * Task and planning tools * Web fetch and search Permissions are scoped and audited. ## Setup scripts (Optional) ### Repository setup script You may add a `cyrus-setup.sh` file to your repository root. This script runs when Cyrus creates a new worktree and is useful for: * Installing dependencies * Bootstrapping environments * Running checks See: [Setup Scripts](/setup-scripts) ### Global setup Team-wide setup is handled automatically and does not require configuration. ## Your existing Claude Code setup Cyrus runs Claude Code inside a Git worktree of your repository, so anything you've already committed under the repo's `.claude/` directory is automatically present in the session and used as-is. No extra wiring, no migration step. The following carry over without any configuration on your part: * **`CLAUDE.md`** at the repo root: project-level instructions the agent reads at the start of every session. * **`.claude/agents/`**: your custom subagents. * **`.claude/commands/`**: your custom slash commands. * **`.claude/settings.json`**: project-scoped settings, hooks, and permissions. If your team has invested in a `.claude/` setup already, keep it. Cyrus picks it up. For the Cyrus-native layer that complements (does not replace) your repo's `.claude/` setup, see [Skills](/skills). ## Design philosophy If something isn’t configurable, it’s intentional. Cyrus optimizes for: * Predictable behavior * Minimal setup * Fewer footguns As new configuration becomes user-relevant, it will be documented here. ## What’s next If something isn’t behaving as expected Get support on Discord