> For the complete documentation index, see [llms.txt](https://doc.wearepatchworks.com/product-documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.wearepatchworks.com/product-documentation/ai/patchworks-ai-studio/platform-agents/health-monitor-agent.md).

# Health Monitor Agent

The Health Monitor Agent finds out why your flows are failing.

{% hint style="success" %}
The Health Monitor Agent is **active** on your account. Access it from the AI Studio or from the Process Flow Run Logs page
{% endhint %}

It investigates your runs, logs, and scripts, explains the root cause in plain language, and can hand the fix straight to the Implementation Agent to apply for you.

Rather than digging through run logs and error groups yourself to work out which step broke and why, you get a single, direct diagnosis - and, when you want it, a fix carried out on your behalf.

<figure><img src="https://2440044887-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FLYNcUBVQwSkOMG6KjZfz%2Fuploads%2FEsIhtwkXU2nrAvURGaR6%2Fhealth-monitor-root-cause.png?alt=media&amp;token=3fb0943a-0c17-47df-83b5-163d8bac26ff" alt=""><figcaption></figcaption></figure>

### What it does <a href="#sources" id="sources"></a>

When you ask about a failure, the Health Monitor Agent reads across the flow's runs, logs, and scripts and synthesises a single answer covering:

* **The failure** - which flow failed, when, how many times, and whether it is still recurring.
* **Root cause** - the underlying reason, separated from the downstream errors it triggered.
* **What to investigate next** - where to look, including practical constraints such as payload retention.
* **A suggested fix** - a concrete change to the flow, which it can apply through the Implementation Agent.

Because it distinguishes the root cause from its knock-on errors, you get the one thing that actually needs fixing rather than a wall of retry and "flow run failed" messages.

<div data-with-frame="true"><figure><img src="https://2440044887-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FLYNcUBVQwSkOMG6KjZfz%2Fuploads%2FhfCVN8VdPqL3cGLyOuP6%2FScreenshot%202026-08-25%20at%2015.39.03.png?alt=media&amp;token=3c16c7c4-c437-4583-b33a-5e0b2eb1d7d8" alt=""><figcaption></figcaption></figure></div>

### How to use it <a href="#sources" id="sources"></a>

The Health Monitor Agent is available from the **AI Studio** page under **Platform Agents**, and can also be opened from the **Process Flow Run Logs** page while you are reviewing failures.

{% stepper %}
{% step %}

### Open the Health Monitor

Navigate to **AI Studio → Platform Agents** and select the Health Monitor card, then click **Start conversation**. Alternatively, open it from the **Process Flow Run Logs** page and select **Health Monitor** in the assistant, so you can ask about failures while looking at them.
{% endstep %}

{% step %}

### Ask what is failing

Ask in plain language - for example, "What are the most recent flow failures?" The agent locates the failing runs, reads the logs and scripts, and returns the root cause with the surrounding detail.
{% endstep %}

{% step %}

### Review the suggested fix

Where there is a sensible remediation, the agent describes exactly what it would change and why, then asks whether you would like it applied.&#x20;
{% endstep %}

{% step %}

### Let it apply the fix

Confirm, and the Health Monitor hands the change to the **Implementation Agent**, which makes it in the flow for you - for example, inserting a validation step ahead of the step that was failing. The agent confirms what was changed and what to expect going forward.
{% endstep %}
{% endstepper %}

{% hint style="info" %}
A fix applied by the Health Monitor is a guard against the symptom, not always a cure for the cause. If a bad payload or upstream issue produced the failure, it is still worth tracing back to the source system or step that caused it.
{% endhint %}