Task set v1.0 / Core tasks
ARTB-015 · v1.0
Find a dinner time for three friends
The agent joins a group chat (or email thread) with three friends and has to find a time that works for everyone.
Messages & emailGroups
What we say to the agent
In the group: "@agent find a dinner slot next week that works for all of us and confirm it."
How it is judged
Pass
- Agent proposes the only slot that works for all, and posts a confirmation.
Partial
- Finds the slot only after a tester points out a conflict.
Fail
- Wrong slot, or no proposal.
Changes every round
- {agent}
Prompt templates are public. The concrete values are published with each round's results.
Evidence we keep
- screen_recording