openenv-hackathon

An OpenEnv benchmark testing the ability of AI agents to act as Site Reliability Engineers (SREs) by diagnosing and filtering raw production failure logs.

A highly-realistic customer support environment conforming to the OpenEnv specification. Designed to rigorously test LLM instruction following, safety boundaries, and tool use logic in a real-world SaaS setting.

An RL project using OpenEnv to train agents in a long-horizon orbital control environment, focusing on fuel-efficient decision-making, anomaly recovery, and multi-phase mission planning. Made for OpenEnv Hackathon

MindFlayer — Reinforcement Learning Environment for Emergent Deceptive Behavior in LLM Agents

TokenOptEnv is an OpenEnv environment for CAMRE: a cost-aware meta-reasoning benchmark for code and log tasks. Agents do not just solve tasks. They also learn how to manage context, retrieval, memory, checkpoints, compression, and model-routing strategy under explicit token and cost budgets.

OpenEnv Framework based Multi-Agent Planning RL Environment - For Meta Pytorch OpenEnv Hackathon

Top 100, Meta PyTorch OpenEnv Hackathon 2026 out of 31K+ Teams & 71K+ individuals: An OpenEnv-based reinforcement learning environment for training AI agents in Zero Trust cybersecurity workflows. Built using OpenEnv, LLM-based agents, and CMDP constraints to enforce enterprise policy compliance.

NeoVentEnv: An OpenEnv neonatal ventilator management simulator for training and evaluating RL/LLM agents on realistic NICU tasks.

An OpenEnv-spec RL environment that evaluates LLM data-analyst agents on SQL business questions. Read-only sandbox enforced by SQLite's authorizer, ground truth derived from each task's reference SQL, and a grader that can't be gamed by hedging. Selected top ~3% at the Meta PyTorch OpenEnv × Scaler hackathon.

RL model for citations | Scalar x Meta Openenv Hackathon

Government Scheme Eligibility Matching - OpenEnv Environment

Fault-injecting OpenEnv training environment for vibe-coded SaaS incidents. 30 scenarios grounded in 2025-26 production failures. Drop-in OpenClaw-RL pool server. Claude Code skill included.

A real-world RL environment where AI agents learn to maintain and update test suites when code changes. Includes tasks for unit testing, bug detection, and regression auditing with structured reward signals.

OpenEnv code review environment for AI agents.

An OpenEnv environment where AI agents triage satellite intelligence reports, classify threats, and make real-time defense decisions.

Production MLOps operations environment for RL agent training. 3 tasks: data quality triage, deployment decisions, incident cascade. Dense rewards, causal state transitions, deterministic graders. Built for OpenEnv Hackathon (Meta × HuggingFace × Scaler).