Monitoring Agent Runs and Logs

Track what ran. See what worked. Spot what didn’t.

Every time an Agent runs, whether manually, on a schedule, or as part of a chain, it leaves a record. You can review these logs to verify what was generated, when it ran, what inputs were used, and whether anything failed. This transparency is especially useful when running multiple Agents across sites, campaigns, or teams.

Where to Monitor Agent Activity

1. Content Created Tab

Your first stop for outputs. This shows the latest content items created by any Agent.

Each entry includes:

Agent name
Time and date of execution
Output preview (clickable)
Links to the original Agent or chain
Tags or labels (e.g., site, campaign, voice)

2. Agent Run History

Accessible from the Agents section. Click on any Agent to view its full run log.

You’ll see:

A chronological list of all runs for that Agent
Whether each run was manual, scheduled, or chained
Input fields used for that run
Any Brand Voice or tags applied
A link to the resulting content (or error)

Error Handling and Logs

When an Agent fails:

A warning icon will appear in the run log
The error message will show what went wrong (e.g., missing input, source not found, formatting issue)
You can click Retry to fix and re-run with corrected inputs

Common causes of Agent failure:

Invalid or empty input fields
Datastream returns no usable content
Source article URL is broken or inaccessible
Exceeded content or character limits

Log Filtering and Search

Quickly locate runs using:

Date range filters
Agent type (e.g., “Quote Verifier”)
Run status (Success, Failed, Scheduled, Canceled)
Tags or Labels (Campaign, Brand Voice, Site)

Use Cases

QA Review: Spot check multiple article drafts from a campaign
Debugging: Investigate failed quote verification runs
Team Oversight: See who ran what, when, and with which instructions
Audit Trail: Verify how content was produced, including inputs, timing, and method

Tip: Link Logs to Agent Templates

For repeat use cases, you can open a successful run, duplicate the Agent with inputs intact, and adjust as needed. This allows teams to build reliable templates over time while learning from past runs.