Spaces:
Sleeping
Sleeping
| name: support_ticket_env | |
| version: "1.0.0" | |
| description: > | |
| A real-world customer support ticket triage environment. | |
| An AI agent acts as a support executive: it classifies incoming tickets, | |
| selects the correct action (reply / escalate / close), and resolves | |
| multi-ticket queues efficiently. | |
| author: AlgoCore | |
| tags: | |
| - openenv | |
| - customer-support | |
| - triage | |
| - nlp | |
| - real-world | |
| tasks: | |
| - id: 1 | |
| name: Classification | |
| difficulty: easy | |
| description: > | |
| Given a customer ticket, predict the correct category | |
| (billing | technical | account | general | refund). | |
| score_range: [0.0, 1.0] | |
| - id: 2 | |
| name: Action Selection | |
| difficulty: medium | |
| description: > | |
| First classify the ticket, then choose the best action: | |
| reply, escalate, or close. | |
| score_range: [0.0, 1.0] | |
| - id: 3 | |
| name: Full Resolution | |
| difficulty: hard | |
| description: > | |
| Handle a queue of 3 tickets. For each ticket classify it, | |
| choose the right action, and (if replying) craft a relevant reply. | |
| Bonus for fewer steps. | |
| score_range: [0.0, 1.0] | |
| action_space: | |
| type: SupportAction | |
| fields: | |
| action_type: "classify | reply | escalate | close" | |
| category: "billing | technical | account | general | refund (required for classify)" | |
| reply_text: "string (required for reply)" | |
| reason: "optional justification" | |
| observation_space: | |
| type: SupportObservation | |
| fields: | |
| ticket_id: string | |
| ticket_text: string | |
| task_id: integer | |
| current_category: "string | null" | |
| resolved: boolean | |
| step_count: integer | |
| feedback: string | |
| score: float | |
| docker_image: "support-ticket-env:latest" | |
| hf_space: "AlgoCore/support-ticket-env" | |