> ## Documentation Index
> Fetch the complete documentation index at: https://docs.upsolve.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# API Activity

> A log of every change your agent, your rules and MCP clients made through your APIs: who it was for, what was called and what your API answered.

Once you connect your own APIs with [API connectors](https://docs.upsolve.ai/ai-agent-builder/interact/api-connectors.md), the agent and your users' rules can make changes in your product: flag a transaction, update a record, create a ticket. **API activity** is the record of every one of those changes, so you can answer "who changed that, and what did the API say?"

Open it from **Workspace settings → API activity**.

![Workspace settings, API activity tab: the 30-day summary, the Connector, User, Operation, Outcome and Source filters, Export CSV, and the list of writes (empty here)](https://mintcdn.com/upsolve/Nurc8hPkBODmwDS1/images/ai-agent-builder/workspaces/api-activity.png?fit=max\&auto=format\&n=Nurc8hPkBODmwDS1\&q=85\&s=76b91a2685f50ff97661899d7863d9ff)

## What is recorded

Every **write** sent to one of your APIs from this workspace, from three sources:

* **Chat**: a write the agent prepared in a conversation and the person confirmed. Writes the person **cancelled** are recorded too.
* **Rule**: a rule whose action calls your API.
* **MCP client**: a call from Claude or another MCP client.

Reads aren't recorded, and neither are the arguments, request bodies or responses: only what was called and what came back. Rows are kept for 90 days.

## The summary

The strip at the top covers the last 30 days: how many writes were sent (and how many cancelled), what share of them failed, and the three operations called most.

## The table

| Column | What it shows |
| - | - |
| **Time** | When the write was sent. |
| **Source** | **Chat**, **Rule** or **MCP client**. |
| **Actor** | The person it was sent for. |
| **Connector** | Which API. |
| **Request** | The method, the path that was called and the operation's name. |
| **Status** | The HTTP status your API answered with. |
| **Latency** | How long your API took. |
| **Outcome** | **Succeeded**, **Refused by the API**, **Connection lost**, **Failed, retryable** or **Cancelled**. |
| **Confirmed** | **Yes** if a person confirmed it, **Cancelled** if they didn't, and a dash where no confirmation applies (a scheduled rule, for example). |
| **Chat** | **Open chat** goes to the conversation the write came from. |

Narrow the list with the **Connector**, **User**, **Operation**, **Outcome** and **Source** filters, and use **Load more** to go further back.

## Export

**Export CSV** downloads the rows matching your filters, up to the newest 5,000.

## Who can see it

Owners, admins and members of the workspace. Read-only members and end users can't.

## Next steps

> [card] [API connectors](https://docs.upsolve.ai/ai-agent-builder/interact/api-connectors.md)
>
> Connect your API and choose which endpoints the agent may call.

> [card] [Rules & Actions Monitoring](https://docs.upsolve.ai/ai-agent-builder/rules-monitoring.md)
>
> Every rule in the workspace and every run.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.