Curia is your AI executive office — a team of specialized agents that reads your email, manages your calendar, tracks expenses, conducts research, and responds on your behalf across Signal and email. Unlike typical AI tools, Curia runs on your own server, maintains a persistent knowledge graph of your world, and enforces governance and audit trails that enterprise security demands.Documentation Index
Fetch the complete documentation index at: https://docs.meetcuria.com/llms.txt
Use this file to discover all available pages before exploring further.
Quick Start
Get Curia running and send your first message in under 15 minutes.
Core Concepts
Understand agents, skills, memory, channels, and the autonomy engine.
Configuration
Customize LLM providers, email accounts, autonomy settings, and more.
API Reference
Integrate with Curia over HTTP — send messages and stream real-time events.
What Curia does
Curia handles the operational work that consumes an executive’s time — automatically, with your voice and your boundaries.Email & Calendar
Reads your inbox, extracts action items, drafts replies, and manages your calendar — all on your behalf.
Research
Multi-session research tasks that build on previous findings, with full web search and page fetching.
Knowledge Graph
Persistent memory of people, decisions, and relationships — with temporal awareness and semantic search.
Autonomy Engine
A single score (0–100) controls how independently Curia acts. Adjust it with a single command.
Why Curia
Your data stays on your server
Curia is single-tenant and self-hosted. Your emails, contacts, and knowledge graph never leave your infrastructure.
Every action is auditable
An append-only audit log traces every event — every message received, tool invoked, and decision made — with full causal chains.
Governance by architecture
Security boundaries are enforced at the framework level, not by policy. A compromised email adapter cannot invoke tools or write to memory — it is architecturally prevented.
Curia is currently in pre-alpha. Core functionality is implemented and working; some advanced features are in progress. See the GitHub repository for current status.