System Architecture
This section covers the core architecture of Evonic AI — the building blocks that power every agent, skill, and plugin.
┌──────────────┐ │ Frontend │ │ (Vanilla) │ └──────┬───────┘ │ HTTP ┌──────▼───────┐ │ Flask App │ │ (app.py) │ └──────┬───────┘ │ ┌──────────────────────┼──────────────────────┐ │ │ │ ┌─────────▼─────────┐ ┌───────▼───────┐ ┌─────────▼─────────┐ │ Agents │ │ Skills │ │ Plugins │ │ │ │ │ │ │ │ • Agent Runtime │ │ • Tool Defs │ │ • Event Handlers │ │ • Sessions │ │ • Backends │ │ • Flask Routes │ │ • Knowledge Base │ │ • Lifecycle │ │ • Config Vars │ │ • Tools │ │ │ │ │ └────────┬──────────┘ └───────┬────────┘ └────────┬──────────┘ │ │ │ ┌────────▼──────────────────────▼──────────────────────▼──────────┐ │ Agent Runtime Engine │ │ │ │ message_received → processing → llm_call → tool_execute → │ │ final_answer → message_sent → turn_complete │ │ │ │ ┌──────────────┐ ┌──────────────┐ ┌──────────────┐ │ │ │ LLM Client │ │ Tool Registry│ │ Event Stream │ │ │ │ (OpenAI API) │ │ (plugins/ │ │ (pub/sub bus)│ │ │ └──────────────┘ │ skills/) │ └──────────────┘ │ │ └──────────────┘ │ └────────────────────────────────────────────────────────────────┘ │ ┌────────▼───────────────────────────────────────────────────────┐ │ Persistence Layer │ │ SQLite (models/db.py) — sessions, messages, agents, models, │ │ evaluations, results, and more │ └───────────────────────────────────────────────────────────────┘Key Components
Section titled “Key Components”| Component | Description | Docs |
|---|---|---|
| Agents | Agent runtime, sessions, knowledge bases, tools, and channels | Overview |
| Skills | Installable packages that bundle tool definitions with Python backends | Skills |
| Plugins | Event-driven extensions that respond to platform events | Overview |
| Models | Model configuration and management | Local Models |
| Events | Platform event stream for plugin integration | Events |
Next Steps
Section titled “Next Steps”- Read the Agents overview to understand the core platform
- Explore Skills to extend agent capabilities
- Learn about Plugins to build custom integrations
- Review Events to understand the event-driven architecture