Spaces:
Sleeping
Sleeping
| name: multi-domain-ai-agent-environment | |
| version: "2.0.0" | |
| description: > | |
| A real-world OpenEnv environment testing AI agents across four professional | |
| domains: Customer Support Triage, Legal Document Review, Clinical Patient | |
| Triage, and Software Engineering PR Review. Each domain contains 3 tasks | |
| progressing from easy classification to hard open-ended generation. | |
| author: OpenEnv Contributor | |
| tags: | |
| - customer-support | |
| - legal | |
| - clinical | |
| - engineering | |
| - nlp | |
| - multi-domain | |
| - real-world | |
| - triage | |
| tasks: | |
| - id: ticket_classification | |
| name: Ticket Classification & Routing | |
| difficulty: easy | |
| score_range: [0.001, 0.999] | |
| domain: customer_support | |
| - id: response_drafting | |
| name: Response Drafting & Quality | |
| difficulty: medium | |
| score_range: [0.001, 0.999] | |
| domain: customer_support | |
| - id: queue_management | |
| name: SLA Queue Management | |
| difficulty: hard | |
| score_range: [0.001, 0.999] | |
| domain: customer_support | |
| - id: multi_turn_conversation | |
| name: Multi-Turn De-escalation | |
| difficulty: very_hard | |
| score_range: [0.001, 0.999] | |
| domain: customer_support | |
| - id: legal_clause_identification | |
| name: Legal Clause Identification | |
| difficulty: easy | |
| score_range: [0.001, 0.999] | |
| domain: legal | |
| - id: legal_risk_flagging | |
| name: Legal Risk Flagging | |
| difficulty: medium | |
| score_range: [0.001, 0.999] | |
| domain: legal | |
| - id: legal_clause_redlining | |
| name: Legal Clause Redlining | |
| difficulty: hard | |
| score_range: [0.001, 0.999] | |
| domain: legal | |
| - id: clinical_triage_classification | |
| name: Clinical Triage Body System | |
| difficulty: easy | |
| score_range: [0.001, 0.999] | |
| domain: clinical | |
| - id: clinical_esi_assignment | |
| name: Clinical ESI Assignment | |
| difficulty: medium | |
| score_range: [0.001, 0.999] | |
| domain: clinical | |
| - id: clinical_triage_note | |
| name: Clinical Triage Note Generation | |
| difficulty: hard | |
| score_range: [0.001, 0.999] | |
| domain: clinical | |
| - id: pr_type_classification | |
| name: PR Type Classification | |
| difficulty: easy | |
| score_range: [0.001, 0.999] | |
| domain: engineering | |
| - id: pr_bug_identification | |
| name: PR Bug Identification | |
| difficulty: medium | |
| score_range: [0.001, 0.999] | |
| domain: engineering | |
| - id: pr_review_comment | |
| name: PR Code Review | |
| difficulty: hard | |
| score_range: [0.001, 0.999] | |
| domain: engineering | |
| endpoints: | |
| reset: POST /reset | |
| step: POST /step | |
| state: GET /state | |
| tasks: GET /tasks | |
| grader: POST /grader | |
| baseline: POST /baseline | |
| observation_schema: app.models.Observation | |
| action_schema: app.models.Action | |
| reward_schema: app.models.Reward |