> For the complete documentation index, see [llms.txt](https://docs.countercyclical.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.countercyclical.io/agents/research-monitors.md).

# Research Monitors

Turn a completed Agent run into recurring research or follow-up work and inspect what changes over time.

A Research Monitor turns a completed Agent run into a recurring Orchestration. Use it to revisit the same coverage question as new evidence becomes available, review changes against a baseline, or follow up on findings that meet a defined threshold.

### Choosing When to Create a Monitor

Create a monitor when someone on the team would otherwise repeat substantially the same research on a regular cadence. Start with an eligible Agent response you can use to establish what the team knows, what should be checked next, and what would merit review.

An event such as a new filing or a Pipeline stage change can also lead to follow-up work. Use a monitor for a scheduled research question. Use the relevant alert or Stage Action when the work should start from a configured event.

### Choosing a Monitor Workflow

| Workflow | When to Use It                                                                                                                                  |
| -------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |
| Research | Recheck an investment question, compare evidence with an earlier observation, and return findings for review.                                   |
| Task     | Repeat work that can propose a workspace change. Each run still follows current permissions and any applicable review or approval requirements. |

### Creating a Monitor

{% stepper %}
{% step %}

#### Open the Completed Run

Open the Agent response and choose **Create Monitor** where available. Name the monitor for the subject and the research question so it is easy to recognize later.
{% endstep %}

{% step %}

#### Review the Recurring Objective

Choose **Research** or **Task**. Review the starting objective and available baseline context. Refine the scope, timeframe, sources, and criteria for a material change. Check any references and skills selected for the recurring work.
{% endstep %}

{% step %}

#### Review and Enable the Schedule

Choose **Review Schedule**. Confirm the timing, time zone, and relevant dates, then enable the schedule when you are ready. Reopen the monitor to check the next run.
{% endstep %}
{% endstepper %}

{% hint style="warning" %}
**A New Monitor Starts Paused**

Creating the monitor saves it for review. The workflow begins recurring execution after you enable its schedule.
{% endhint %}

### Writing a Recurring Objective

Describe the evidence the Agent should check on each run and the period it should compare with the baseline. Replace temporary phrases such as “this week” with a time window that makes sense on future runs. State what constitutes a material change and what you expect the Agent to report when the evidence is incomplete.

The setup can use earlier findings as a reference point. Instruct the Agent to look for fresh evidence on each run. If the original conversation used files, explicitly review which sources the recurring workflow needs. Attachments from that conversation are not automatically carried into scheduled runs.

### Following and Managing a Monitor

Return to the linked monitor from the conversation or open the resulting Orchestration. Use the run history and **Inspect Run** to see which steps executed, review available evidence and results, and handle any pending decision. Review the outcome alongside the objective before treating a reported change as a change to your investment thesis.

To adjust the cadence or stop future starts, open the schedule from the orchestration. Check whether any run has already started before making a change. If an expected run is missing, check the schedule state, the previous run, current permissions, and any model, credit, or approval requirements shown in the product.

#### Related Pages

{% content-ref url="/pages/Ff3W2jW4IMpZKEEf5yRi" %}
[Conversations](https://docs.countercyclical.io/agents/conversations.md)
{% endcontent-ref %}

{% content-ref url="/pages/IOdV59IPEnBdD3TcI7Zg" %}
[Adding Sources to an Agent Request](https://docs.countercyclical.io/agents/adding-sources-to-an-agent-request.md)
{% endcontent-ref %}

{% content-ref url="/pages/U8A8RPCDt7CQ2VNFGySE" %}
[Runs](https://docs.countercyclical.io/agents/orchestrations/runs.md)
{% endcontent-ref %}

{% content-ref url="/pages/x5NcwZqAn0tjF3KR1AXS" %}
[Inspection](https://docs.countercyclical.io/agents/orchestrations/inspection.md)
{% endcontent-ref %}

{% content-ref url="/pages/SCKkJzOQmkavrgSEVNVK" %}
[Stage Actions](https://docs.countercyclical.io/workspaces/stage-actions.md)
{% endcontent-ref %}
