> For the complete documentation index, see [llms.txt](/llms.txt).
> A full single-fetch corpus is available at [llms-full.txt](/llms-full.txt).
---
title: Run
description: Invoke your deployed workflow — HTTP, SDK, schedule, or Studio.
page_type: tutorial
audience: both
last_verified: 2026-05-13
---

This is stage **3 of 5** of [the AGNT5 loop](/docs/get-started/loop.md). Once your workflow is deployed, trigger it from wherever the work originates.

**From Studio** — pick the workflow, enter input, click **Run**. Same UX as `agnt5 dev`, against the managed environment.

**From the CLI:**

```bash
agnt5 run digest --input '{"limit": 5}'
```

**From your app — Python SDK:**

```python
from agnt5 import Client

client = Client(gateway_url="https://api.agnt5.com", api_key="agnt5_sk_...")
result = client.run("digest", {"limit": 5})
```

**Over HTTP:**

```bash
curl -X POST https://api.agnt5.com/v1/workflows/digest/run \
  -H "Authorization: Bearer agnt5_sk_..." \
  -H "Content-Type: application/json" \
  -d '{"limit": 5}'
```

Each invocation gets its own trace in Studio — head to [Observe](/docs/get-started/observe.md) next.