loop-guard

A lightweight, framework-agnostic loop detection library for LLM agents. Detects repeated tool calls, identical inputs, unchanged outputs, and stalled execution before they become infinite loops. Returns actionable recovery suggestions and integrates into any Node.js AI agent with just two function calls.

Detect & break thinking-chain loops in DSH agents at the turn boundary (no proxy). Ported from ollama-loop-guard.

GLM Mode: fully tuned coding agent preset for zhipuai/glm-5.3-flash in DeepSeek Harness — PTC presentation, guard family, isolated compaction

DSH plugin that stops a stalled agent: repeats output detection, near-duplicate tool call guard, optional runaway brakes.

A project-agnostic multi-agent collaboration protocol for the DeepSeek Harness: loop guards that stop runaway agent loops, a six-field handoff format, risk-tiered review routing, and a fixed verify→commit→push→review→close sequence. One skill instead of the same rules copy-pasted into every agent's instructions.

Loop-guard + pipeline-cache patch untuk KytyPS5 (github.com/KytyPS5/KytyPS5) — ASTRO's Playroom PPSA01325 di Linux Xubuntu: astro enter game mode 2-3fps.. sekarang bisa main; semua edit oleh AI DeepSeek V4.1 Flash

Hermes Agent plugin: LSP semantic code navigation tool (definitions/references/hover/symbols/diagnostics), fail-closed bubblewrap sandbox tool, and advisory loop-guard. Ported from DeepSeek Harness architecture. AI-agent-native, zero core changes.

多智能体监工：一个独立模型实时审阅推理文本与工具调用，检测死循环 / 空转 / 偷懒，并按 WARN / BLOCK / STOP 三档落地。DSH 宿主插件 + 浏览器设置界面。

A deterministic guard for repeated agent actions, iterations without progress, unresolved red flags and unsupported completion claims.

DSH 插件：检测并打断 agent 循环中的纯文本重复退化（generation degeneration）。7917 条真实消息实测，误报率 0.03%。DeepSeek Harness plugin that detects and breaks pure-text degeneration loops in agent output.

DSH 循环守卫：单轮工具调用上限、重复动作识别、长回合沉默播报。Caps tool calls per turn, detects repeated identical actions, and speaks up when a long turn has gone quiet without addressing the user.

Boost LLM reliability with dynamic sampling, retry logic, and a lightweight Go‑based proxy.