Core Concepts

Runiq.Net is a code-first agent runtime for .NET applications.

It gives ASP.NET Core teams a native way to define AI agents in C#, attach strongly typed tools, stream model responses, connect reusable context sources, orchestrate workflows, and inspect runtime activity through an embedded dashboard.

Runiq is designed for developers who want more than a direct LLM API call. In a real product, an AI feature usually needs application instructions, typed business capabilities, source-backed context, predictable execution paths, streaming events, and a way to inspect what actually happened. Runiq brings those pieces into your ASP.NET Core host instead of asking you to move your AI logic into a separate service or glue together unrelated tools.

Runtime and dashboard

With Runiq, your app owns the runtime. Agents, tools, Context Spaces, and workflows are registered in the same ASP.NET Core application that owns your services and configuration.

The embedded dashboard is served from that host with UseRuniqDashboard. It gives developers a place to inspect registered agents, test conversations, review runtime activity, run tools, browse Context Space documents, and inspect workflow execution.

Agents

An agent is a registered AI runtime component inside your .NET application. It has a stable id, display name, instructions, model/provider configuration, optional runtime behavior settings, attached tools, optional Context Spaces, and dashboard visibility.

Agents should stay focused. The repository samples use separate agents for responsibilities such as weather analysis, place suggestions, final travel planning, and expense data analysis.

Tools

A tool is a normal typed C# class that can call your services, repositories, APIs, databases, validators, or domain logic.

Tools make application capabilities explicit. An agent can only call the tools attached to it, and dashboard runs show tool activity separately from model output.

Context Spaces

A Context Space gives agents source-backed knowledge such as product docs, policies, runbooks, markdown files, PDF guides, or sample-specific domain documents.

Context search is different from a tool call. Context search retrieves relevant text from attached sources. A tool call executes typed .NET code.

Workflows

A workflow lets the application own the order when a feature needs deterministic multi-step execution.

In the workflow travel planner sample, the application defines the sequence weather -> places -> planner. Each step runs an agent, but the workflow transition rules live in C# instead of being hidden in a prompt.

Why this exists for .NET applications

Runiq makes AI behavior part of the application architecture. You can keep business rules in C#, keep dependency injection, keep existing service boundaries, and expose only the capabilities an agent should be allowed to use.

You can start with one focused agent, test it in the dashboard, then add tools, Context Spaces, or workflows only when the product behavior needs them.