Update agent selection for HF-hosted and local LLM paths

Finalize RL-first environment and explorer UI

Reframe FraudShield as enterprise FraudOps workflow

Harden observation leakage and calibrate baseline

Upgrade baseline with agentic investigation policy

Add agentic investigation workflow and runtime metadata

Use injected proxy credentials for baseline agent

Fix: Fallback to heuristic when only one LLM env var is set; improve error handling

Support HF-safe Space variable names

Finalize OpenEnv submission readiness

Prepare FraudShield for OpenEnv submission

Initial commit - FraudShield OpenEnv environment