This article covers edge cases and unexpected behaviors that can cause automatic agent triggers not to fire or to behave differently than expected. For setup instructions, see Configure agent trigger settings. For an overview of when each trigger type fires, see Agent trigger types.
⠀
Configuration warnings and errors
Warning: Agents of the same type on the same sheet and trigger
⠀
This warning appears when two or more agents of the same type (for example, two General Testing Agents) are assigned to the same sheet and trigger. The full warning message reads: Agents marked with a warning icon are of the same type and are assigned to the same sheet and trigger. Only one agent will be chosen to run on the selected sheet. When the trigger fires, only one agent runs. The most recently added trigger configuration for that agent type is used as the tiebreaker.
⠀
To resolve this: Remove the duplicate configuration(s) from the Trigger Configurations menu in Engagement Settings > Agents.
⠀
Error: Configurations reference deleted agents or sheets
⠀
This error appears when a trigger configuration references an agent pipeline or sheet that no longer exists.
⠀
To resolve this: Remove any configurations that reference deleted agents or sheets from the Trigger Configurations menu in Engagement Settings > Agents.
⠀
The trigger didn't fire
A run is already in progress on that row
If a Fieldguide-triggered run is already active on a row, a second auto-trigger for that row is skipped — the new run won't queue behind it.
⠀
If the active run has been running for more than 30 minutes, Fieldguide assumes it's stalled and releases the block automatically. The next trigger event can then proceed normally.
⠀
A trigger type isn't enabled for your firm
Each trigger type can be enabled or disabled for your firm independently. If a trigger you've configured isn't firing, that trigger type may not be enabled for your firm. Contact Fieldguide support to verify which trigger types are active.
⠀
The trigger fired, but results were unexpected
The agent ran on the wrong column
When a trigger fires, the agent automatically selects which column to run against. This selection can't be overridden for automatically triggered runs. The ability to specify a column for auto-triggered runs is on Fieldguide's feature roadmap.
⠀
If the agent is consistently running on the wrong column, check your column names. Auto-triggered runs select a column using this priority:
Columns whose title contains "test procedures" or "testing procedures" — Agent Test Procedures is selected first if it exists and contains data
Columns whose title contains "test," "testing," or "tests"
Alphabetical order as a fallback
For column naming best practices, see Field Agents for Testing.
⠀
Bulk uploads caused multiple runs on the same row
Uploading many documents to the same row at once can produce more auto-triggered runs than expected. When multiple documents are uploaded in rapid succession, Fieldguide may not be able to consolidate them into a single run.
⠀
If this happens, you can delete extra runs from the agent side panel.
⠀
A trigger fired while a reviewer was actively working
If a new auto-trigger fires while a reviewer is working in the agent panel, the new run can overwrite any edits the reviewer has already made in the agent panel.
⠀
You can prevent this from happening by having the preparer sign off on sheet rows: once a preparer signs off on a row, all auto-triggers for that row are paused until the sign-off is lifted. If a reviewer needs uninterrupted time to review agent output, have the preparer sign off on the row before the review begins.
⠀
Delays and external factors
Auto-triggered runs are delayed
Auto-triggered runs queue behind manual runs. If your firm has a high volume of manual runs in progress, auto-triggered runs may be delayed by a few minutes. If you suspect a delay, check whether manual runs are active in the same engagement — you can view active runs in the agent panel. Auto-triggered runs will resume once the manual queue clears.
⠀
An LLM outage produced a false result
During outages from AI providers, an agent run can complete but return a false "no exceptions noted" result instead of failing with an error. If a completed run looks suspiciously clean, rerun the agent manually to verify the result.
⠀
Document issues
The agent didn't process an uploaded document
If an agent run completed but the agent doesn't appear to have referenced a document you uploaded, the file may be in a format that agents can't process. When a run encounters an unsupported file type — such as a ZIP archive — the run completes normally but silently skips that document.
To check which file formats are supported, see File types supported by Field Agents for testing.
A very large file caused the run to stall or error
If an agent run has been stuck in "in progress" for an unusually long time, or failed with an error, a very large document on the row may be the cause. Files with very high page counts or very large spreadsheets — such as an Excel workbook with millions of rows — can exceed the processing limits of the agent.
To investigate, check the documents currently on the row. If any appear unusually large, try removing them and rerunning the agent. If the run completes after removing a document, that file is likely the cause.