# Harness

> Define a loop, assemble a Rig, run a session, observe its events, and shut it down cleanly with the Harness runtime.

- Path: `Modules > Harness - Agent Runtime`
- Human: https://looprig.com/docs/modules/harness
- Machine index: https://looprig.com/llms.txt

## Repository

| Field | Value |
| --- | --- |
| Repository | `github.com/looprig/harness` |
| Version | `v0.28.0` |
| GitHub | [looprig/harness](https://github.com/looprig/harness) |

## Description

Define a loop, assemble a Rig, run a session, observe its events, and shut it down cleanly with the Harness runtime.

## Where it fits

Harness is useful on its own when an application needs to turn model calls and tools into a session-oriented agent runtime. Within Looprig, it composes [Inference](/docs/modules/inference.md), [Storage](/docs/modules/storage.md), gates, workspaces, delegation, and tools behind Loops and Rigs. Harness owns runtime coordination; applications still choose models, authority, persistence, and interfaces.

## Dependencies

- [Core](/docs/modules/core.md)
- [FSStore](/docs/modules/fsstore.md)
- [Inference](/docs/modules/inference.md)
- [Storage](/docs/modules/storage.md)

## Dependents

- [ACP](/docs/modules/acp.md)
- [Classifiers](/docs/modules/classifiers.md)
- [Client](/docs/modules/client.md)
- [Foreign Loops](/docs/modules/foreignloops.md)
- [MCP](/docs/modules/mcp.md)
- [Tools](/docs/modules/tools.md)
- [TUI](/docs/modules/tui.md)
- [Workflows](/docs/modules/workflows.md)