# Curia

## Docs

- [Create a custom agent: YAML config and handlers](https://docs.meetcuria.com/agents/building-custom-agents.md): Define a specialist agent in YAML, configure its model, memory scopes, and error budgets, and optionally add a TypeScript handler for custom logic.
- [Built-in agents](https://docs.meetcuria.com/agents/built-in-agents.md): The agents Curia ships with — what each does, when it's invoked, and how the coordinator dispatches to specialists.
- [Agents: Curia's coordinator and specialist AI model](https://docs.meetcuria.com/agents/how-agents-work.md): The Coordinator agent handles every message while specialist agents work behind the scenes — the sender never knows multiple agents are involved.
- [Building custom channels](https://docs.meetcuria.com/channels/building-custom-channels.md): How to create a new channel integration — implementing a channel adapter, declaring trust levels, and registering with the message bus.
- [CLI channel](https://docs.meetcuria.com/channels/cli.md): Using the command-line interface to interact with Curia locally — the simplest and highest-trust channel.
- [Configure Curia's email channel with Nylas integration](https://docs.meetcuria.com/channels/email-setup.md): Configure Nylas to give Curia a dedicated email account for sending and receiving, and optionally monitor your own inbox in observation mode.
- [Channels: email, Signal, CLI, and HTTP API support](https://docs.meetcuria.com/channels/how-channels-work.md): Email, Signal, CLI, and HTTP are thin adapters with trust levels. Curia normalizes all messages into a common format and gates actions by channel trust.
- [HTTP channel](https://docs.meetcuria.com/channels/http.md): Interact with Curia programmatically via REST API and server-sent events.
- [Connect Signal for encrypted messaging with Curia](https://docs.meetcuria.com/channels/signal-setup.md): Connect a phone number to Curia via signal-cli for end-to-end encrypted messaging with your agents — Curia's highest-trust communication channel.
- [Architecture](https://docs.meetcuria.com/core-concepts/architecture.md): How a message flows through Curia — from inbound channel to coordinator dispatch, agent execution, skill invocation, and outbound delivery.
- [Autonomy](https://docs.meetcuria.com/core-concepts/autonomy.md): One number from 0 to 100 controls how independently Curia acts, across all channels and skills.
- [How Curia verifies contacts and authorizes senders](https://docs.meetcuria.com/core-concepts/contacts-trust.md): Curia verifies every sender before acting on their behalf, using trust scores, channel authentication, and deterministic permissions that you control.
- [Dreaming](https://docs.meetcuria.com/core-concepts/dreaming.md): How Curia's dream engine runs background maintenance — knowledge graph decay, confidence updates, and autonomy score adjustments.
- [Curia: governance-first AI executive assistant](https://docs.meetcuria.com/core-concepts/introduction.md): What Curia is, how it works, and why it was built differently — governance by architecture, persistent memory, and an autonomy engine you control.
- [The layers model](https://docs.meetcuria.com/core-concepts/layers-model.md): How Curia's governance layers compose — from channel trust to dispatch policy, coordinator autonomy checks, and skill-level action risk.
- [Memory and the knowledge graph](https://docs.meetcuria.com/core-concepts/memory.md): How Curia stores knowledge across sessions — four memory tiers, temporal confidence, semantic search, and the knowledge graph browser.
- [Releases and roadmap](https://docs.meetcuria.com/core-concepts/releases.md): Where to find Curia's release history, changelog, and planned milestones.
- [Schedule recurring and one-shot tasks in Curia](https://docs.meetcuria.com/core-concepts/scheduling.md): Set up recurring cron jobs in agent YAML, create one-shot reminders via conversation, and understand how long-running tasks survive restarts.
- [Initial configuration](https://docs.meetcuria.com/get-started/configuration.md): Set up environment variables, LLM providers, email, and Signal for your Curia deployment.
- [Deployment and self-hosting](https://docs.meetcuria.com/get-started/deployment.md): Get Curia running on your own server — prerequisites, database setup, configuration tiers, and first-run verification.
- [Set up Curia from minimum to full configuration](https://docs.meetcuria.com/get-started/installation.md): Clone, configure, and start your Curia instance in three tiers — minimum setup gets the CLI working; full setup adds email, search, and Signal messaging.
- [Playbooks and best practices](https://docs.meetcuria.com/get-started/playbooks.md): Suggested workflows and recipes for common Curia use cases — email triage, weekly briefings, contact management, and more.
- [Agent manifest schema](https://docs.meetcuria.com/references/agent-manifest-schema.md): Complete YAML schema reference for Curia agent definitions — all fields, types, defaults, and configuration options.
- [Channel configuration schema](https://docs.meetcuria.com/references/channel-config-schema.md): YAML schema reference for Curia channel configuration — email accounts, outbound policies, and observation mode.
- [Configuration reference](https://docs.meetcuria.com/references/configuration.md): Complete reference for all Curia configuration options in config/local.yaml — rate limits, security rules, knowledge graph decay, intent drift, and more.
- [Curia HTTP API — overview, auth, and all endpoints](https://docs.meetcuria.com/references/http-api.md): The Curia HTTP API is a REST and SSE interface for programmatic access. Authenticate with a Bearer token; the base URL defaults to http://localhost:3000.
- [GET /api/health — instance health check endpoint](https://docs.meetcuria.com/references/http-api-health.md): Returns the live status of your Curia instance including database connectivity, registered agents, and loaded skills. No authentication required.
- [POST /api/messages — send a message to your agent](https://docs.meetcuria.com/references/http-api-messages.md): Send a message to the Coordinator agent and receive a synchronous response. Supports conversation threading and returns the agent reply within 120 seconds.
- [GET /api/messages/stream — real-time event stream](https://docs.meetcuria.com/references/http-api-stream.md): Open a Server-Sent Events connection to receive agent responses, tool invocations, and tool results in real time, optionally filtered by conversation.
- [Memory schemas](https://docs.meetcuria.com/references/memory-schemas.md): Schema reference for Curia's knowledge graph — node types, edge types, decay classes, sensitivity levels, and temporal metadata.
- [Releases](https://docs.meetcuria.com/references/releases.md): Curia release history, version changelog, and migration notes.
- [Skill manifest schema](https://docs.meetcuria.com/references/skill-manifest-schema.md): Complete JSON schema reference for Curia skill definitions — all fields, action_risk levels, capabilities, and entity enrichment.
- [Curia's append-only audit trail and governance log](https://docs.meetcuria.com/security/audit-log.md): Every Curia event is logged to Postgres before delivery — with full causal tracing across every message, skill call, and agent response for compliance review.
- [Autonomy engine](https://docs.meetcuria.com/security/autonomy-engine.md): How Curia's autonomy score governs skill execution through action_risk levels, audit trails, and planned automatic adjustment.
- [Security and governance in Curia — six layers explained](https://docs.meetcuria.com/security/overview.md): Curia's security model enforces hard architectural boundaries, an append-only audit trail, and exec-grade governance by design — not by policy.
- [Redaction](https://docs.meetcuria.com/security/redaction.md): How Curia redacts sensitive data from outbound messages — built-in PII patterns, per-channel policies, and the redaction pipeline.
- [Build a custom skill or connect an MCP server](https://docs.meetcuria.com/skills/building-custom-skills.md): Create a local skill with a manifest and TypeScript handler, or connect an MCP server — both appear in the skill registry and work identically for agents.
- [Built-in skills](https://docs.meetcuria.com/skills/built-in-skills.md): Catalog of the skills Curia ships with — organized by category, with action_risk levels and descriptions.
- [Skills: how agents take action in the outside world](https://docs.meetcuria.com/skills/how-skills-work.md): Skills connect Curia to the outside world. Each one declares its inputs, outputs, action risk level, and required secrets before the agent can invoke it.