The open-source memory operating system for AI agents. Persistent memory, semantic search, loop detection, agent messaging, crash recovery, and real-time observability.

Runtime for coding agents. Models are getting smarter, but a model is only as capable as the environment supporting it. Atelier is that environment.

Your AI agent just burned $200. AgentGuard stops it at $5. Runtime cost guardrails for AI agents — budget enforcement, loop detection, kill switch. Zero dependencies, MIT licensed.

An open-source cost controller for AI agent loops — stops a loop when it's actually converged and rolls back before it degrades, instead of running to a fixed max_iterations cap. Real-time loop-gain (Aβ) bands + best-so-far rollback. Adapters for LangGraph, CrewAI, AutoGen, LangChain, OpenAI Agents, and Claude Agent SDK; raw API for custom stacks.

UnitCause is a real-time cost governance platform for AI agents. It provides budget enforcement, loop detection, cost analytics, and policy management via a web dashboard, REST API, and Python SDK. Set budgets, stop runaway agents, and monitor spending with ease.

The pytest of agent-loop health: a pip-installable, in-process library that detects reasoning loops, null progress, context pressure, and cost run-away with cheap deterministic checks. No backend, no dashboard, no LLM judge, no data leaving your process — runs in prod and inside training rollouts.

Collapse and persistence in purpose-bearing systems. The observer opens doors.

Behavioral drift detection for LLM agents — catch vocabulary loss, tool changes, loops, and stagnation before they break production.

Stop hallucinated tool calls, break looping behaviour, and keep secrets out of your context window — zero-config guide-rails for Hermes Agent.

Multi-agent reasoning system with loop awareness, human-in-the-loop arbitration, and long-term reasoning memory.

A safety orchestrator for Autonomous Agents. Prevents infinite loops and destructive system calls (rm -rf) by enforcing a "Think-Plan-Act" state machine (H.E.L.M.).

Extreme Low-Bit Inference in Reasoning Models: Failure Modes and Targeted Recovery

Open-source guardrails that auto-kill runaway AI agents — budget enforcement, loop detection, quality verification

Detect when an LLM agent is stuck in a loop (repeats, cycles, no novelty). Zero dependencies.

Real-time AI coding agent loop detector for Claude Code

An active, viability-constrained resource governor for autonomous LLM agents. Prevents token runaway, context-window overflows, and infinite loops without hard crashes.

Loop detector + cost kill switch for LLM agents. Stops runaway agents before they drain your budget.

MiMo Stable — Three-layer defense against degenerate loops in MiMo-series LLMs (xiaomimimo). Empirically validated parameters + language enforcement + runtime detection.

A framework-agnostic, zero-dependency semantic and structural loop detector for AI/LLM agents (LangChain, LangGraph, CrewAI, Autogen). Prevents infinite execution and reasoning oscillation cycles.

Framework-agnostic agent loop detection — sliding window similarity scoring, zero dependencies