Repository navigation
Home
One prompt → a coordinated organization of AI agents that plans, delegates, executes, and verifies work.
Welcome to the community documentation for Cascade AI, a multi-tier AI orchestration platform with model routing, tool execution, and multiple interfaces. This wiki is organized for both end users and developers.
Begin here: Getting-Started · Architecture · CLI-Reference · Cloud-and-Self-Hosting
| I want to… | Wiki page |
|---|---|
| Install Cascade and complete my first task | Getting-Started |
| Understand administrators, managers, and workers | Agent-Orchestration |
| Understand the codebase and components | Architecture |
| Choose models and control spending | Models-and-Budgets |
| Learn CLI commands and the interactive terminal | CLI-Reference |
| Use the Windows/macOS/Linux desktop app | Desktop-App |
| Deploy Cascade Cloud or self-host the web interface | Cloud-and-Self-Hosting |
| Integrate the SDK or OpenAI-compatible API | API-and-SDK |
| Use built-in tools, MCP and connectors | Tools-and-MCP |
| Control a browser or take over an agent session | Browser-Control |
| Create real office files and multimodal outputs | Files-and-Media |
| Review trust boundaries and protect data | Security-and-Privacy |
| Contribute code, run tests or troubleshoot | Development · Troubleshooting |
| Maintain this wiki | Documentation-Maintenance · Weekly-Code-Changes |
| Surface | Purpose | Availability notes |
|---|---|---|
| CLI | Interactive Ink terminal and one-shot runs | Does not provide the live desktop/cloud browser-control interface |
| Desktop | Electron chat, Cockpit, code editor, terminal, integrated browser | Privileged native capabilities require host-side controls |
| Cloud web | Hosted chat, memory, files, account and connector features | Hosted browser requires an operator-configured backend |
| Self-hosted web | Run the cloud application on your own infrastructure | Configure OAuth, storage persistence and TLS before external exposure |
| SDK | Embed orchestration in Node apps | Calling application handles credentials and approvals |
| OpenAI-compatible HTTP API | Integrate existing SDK clients with /v1
|
Exposes Cascade routing modes, not full OpenAI parameter parity |
The pages describe implemented, merged features. Available options vary by interface and configuration. For changes after this baseline, check the repository's main branch, CHANGELOG.md, and Weekly-Code-Changes. Roadmap items and open PRs are not necessarily released features.
Initial baseline: source main commit 352726b, package version 0.82.0. This version is the documented snapshot, not a prediction of the latest available release.