Monitoring | Faustine Docs

Monitoring is the workspace-wide view of activity, logs and errors. It is under Monitoring in the portal's sidebar. For one Web Agent or one gateway, use that object's own logs.

How it works

The page has three views, selected at the top.

View	Shows
Activity	What Faustine did: analyses, updates, versions, tests and recipes.
Logs	What users and agents did: conversations and requests.
Errors	What went wrong: failed runs, failed turns and errored calls.

Activity

Activity is a feed of every section of the workspace, newest first. A run is one row, and the row changes as the run moves from Queued to Running to its result.

Five counters sit above the feed: Running now, Runs, Versions published, Breaking changes and Failed.

Each row has these columns.

Column	Content
When	The time of the last change.
Where	The Web Agent, API or gateway.
What	A sentence that describes the event.
By	The member who started it, or Faustine for automatic runs.
Result	Queued, Running, Completed, Failed, Interrupted, Cancelled or Attention.

Select a row to see its details, including the event type, when it started and the run ID.

Four filters narrow the feed.

Filter	Values
Source	All sources, Web Agents, APIs, Agent Gateway, Faustine Chat
Kind	Everything, Analyses, Updates, Adapter, Versions, Tests, Recipes, Connections, Operations, Source, Access, Environments, Custom domain
Status	Any status, Running, Queued, Completed, Failed, Interrupted, Cancelled
Time	Last 24 hours, Last 7 days, Last 30 days, Any time

The feed loads 50 rows at a time. Select Load more for older ones.

Logs

Logs shows one source at a time, on that source's own table.

Source	Content	Details
Web Agents	Conversations in the widget, across all agents.	Read the logs
Faustine Chat	Conversations of your chat users.	Read Faustine Chat logs and usage
Agent Gateway	Requests to your gateways.	Logs

Errors

Errors keeps a separate list per source, so each error links to the place it came from.

Section	Lists
Failed runs	Analyses, updates, imports and test generations that failed or were interrupted. Open goes to the run.
Web Agents	Turns the widget could not answer, with the message, the model and the HTTP status. View conversation opens it in the agent's logs.
Faustine Chat	Turns the chat could not answer.
Agent Gateway	Requests that errored, in the gateway or in your API.

Use the source selector to show one section, and the time filter to choose Last 24 hours, Last 7 days or Any time. Each list shows up to 20 entries. For gateway requests, Any time covers the last 30 days.

Next steps

Read the logs: conversations of one Web Agent.
Logs: the request log of a gateway.
Choose who gets workspace notifications: get an email when a run fails.
Usage: the tokens these runs and conversations consumed.