Getting Started
Stallari is a local-first personal operations platform. It gives you a Chief of Staff for your own information: chat, daily context, local search, plugin-backed actions, and a clear view of what the local system is doing.
Start with First 30 Minutes. It covers the first run, provider setup, macOS permissions, and where to look when something needs attention.
What To Expect
Section titled “What To Expect”Stallari has three visible surfaces:
| Surface | Purpose |
|---|---|
| Chat | Ask questions, request work, and see privacy routing for each answer. |
| Glance | Read the day: Today, Digests, Search, Insights, and related personal context. |
| Watch | Check system state: Operations for live work, Activity for completed work, and Status for health. |
Settings is for controls. Status is where Stallari shows live state, permission readiness, degraded services, and recovery actions.
First Requirements
Section titled “First Requirements”- A Mac running the current supported macOS release for Stallari.
- A local vault or folder that Stallari can use as its knowledge base.
- At least one model route: local models, a cloud provider, ChatGPT Plus/Pro through the supported OpenAI path, Anthropic, Google, xAI, or a compatible provider configured in Settings.
- Explicit permission for any local private data source you enable, such as Mail indexing.
Safe Defaults
Section titled “Safe Defaults”Stallari does not ask macOS for broad privacy access on first launch. It explains each grant before you trigger it.
Login Items lets the helper run in the background. Full Disk Access may be needed for local corpus features such as Mail indexing. App Management may be needed during update or replacement flows if macOS blocks the action. Plugin and MCP consent is separate: a plugin must be approved for its data classes and action classes before it can use sensitive capabilities.
Continue with First 30 Minutes.
Related concepts
Section titled “Related concepts”- Agency Model — the Chief of Staff mental model behind the surfaces
- Context And Memory — how Stallari decides what an agent sees per task
- Local Vs Cloud — when local models run and when cloud providers run