Needs review
Project:
AI Agents Test
Version:
1.0.x-dev
Component:
Code
Priority:
Major
Category:
Task
Assigned:
Unassigned
Issue tags:
Reporter:
Created:
14 Aug 2025 at 10:09 UTC
Updated:
2 Oct 2025 at 15:21 UTC
Jump to comment: Most recent, Most recent file
Comments
Comment #2
yautja_cetanu commentedComment #3
yautja_cetanu commentedComment #4
blanca.esqueda commentedAssigning this task to me.
I'll give it a try - I'll reach to Bisonbleu or yautja_cetanu for concerns/help.
Comment #5
blanca.esqueda commentedFor reference;
Comment #7
blanca.esqueda commentedComment #9
bisonbleu commentedWiP - Made some progress implementing this using AiAgentStatusPollerService. The latest code in the MR returns a combined Test Group
group_durationas can be seen with a drush command (this is not yet displayed in the UI).I’m now running into some PHPstan issues on GitLab (which I don’t see locally). Looks like they might be linked to how PHPstan is setup in the repo…
Comment #10
bisonbleu commentedCurrently only displaying Total completion time on Test Group Result page (see capture). Please advise where else Test and Test Group completion times should be displayed.
How to test:
Comment #11
yautja_cetanu commentedPerhaps next to Status for each test.
Even better would be a kind of Gantt chart log thing where you can see visually the relative times it takes! But that might be a bit advanced.
Something like: https://claude.ai/public/artifacts/58a06f78-2552-40fc-add8-0eb17f483a02
OR: https://claude.ai/public/artifacts/dd00223a-017f-47c6-9ceb-6c29e741d879
But I think this is a follow-up issue. For now, I think having the time it took for each test specifically next to status would be good.
Comment #12
bisonbleu commented@yautja_cetanu , the examples you provided are really nice… definitely where React shines!
For now I went for the minimalist approach: added a Time column to the Test Group table as you suggested. Also added the Execution Time when viewing a test details.
Comment #13
bisonbleu commentedAfter latest commit, tool_timings data is properly returned.
And in case you're wondering, the "u" in toolu_ is not a typo - it's part of Anthropic's tool use ID format.
When Claude (Anthropic's models) uses tools/functions, it generates unique identifiers for each tool call. The prefix toolu_ stands for "tool use" and is followed by a unique alphanumeric string. This is the standard format that Anthropic's API returns when a model decides to call a tool.