Turn failed AI agent runs into replayable regression tests. Catch regressions before you ship.
grid-gym is an open-source platform for deterministic simulation and validation of smart grid and energy management systems. It supports replay simulations, fault injection, multi-agent control, and Modbus, MQTT, and real-time telemetry for research, development, and integration testing.
Flight recorder, MAST failure attribution & on-call ops loop for AI agents: replay, tamper-evident evidence, triage queue, letter grades, handoff briefs, fleet tail, price catalog. 49 MCP tools. Zero dependencies, Python 3.10+.
Record agent runs. Replay them offline. No cloud, no changes to your agent.
Build, deploy, compare, and independently verify sports-trading agents with sealed evidence, deterministic scoring, and operator-controlled execution.
Offline n8n workflow replay, target-output comparison, and migration readiness gates for autonomous agent transitions.
To associate your repository with the replay-testing topic, visit your repo's landing page and select "manage topics."