跳转至

dsh-llm-fallbacks
类型 智能体 分类 多智能体
星数 ⭐3 状态 ✅ 活跃
能力 context, automation 更新时间 2026-08-15

📖 项目简介

English | 中文 Automatic provider/model fallback chains for dsh (DeepSeek Harness): when an agent's LLM requests keep failing — retries exhausted, auth errors, quota exceeded, rate limiting (429) — the plugin switches provider/model along the fallback chain for the current role, and the current step/turn continues on the target model: tasks are not interrupted by model problems. Install with a single command (see Install): dsh plugin --profile web add ds

✨ 核心特性

  • Automatic fallback for root and subagents: any agent switches down the chain to the next available provider/model on model failure — no
  • Two-block config: block 1 rootChain — the root agent's single fallback chain (empty = root does not fall back); block 2 declared roles
  • Entry syntax: chain entries are provider/model (exact switch) or provider/* (keep the failed model id, switch provider only) — the o
  • Cooldown and revert: models that were switched away from / failed are not re-selected during the cooldown period; `revertPolicy: cooldow
  • Visible behavior: every switch appends a persisted session event fallbacks/switch (from/to/role/reason), alongside info-level logs (ca
  • Safety valves: switching stops and the original error semantics are kept once maxSwitchesPerStep is exceeded for a step, preventing ch

📦 安装

dsh plugin --profile web add dsh-llm-fallbacks   # pin a version with @<version>

🚀 快速开始

npm install dsh-llm-fallbacks   # or: pnpm add dsh-llm-fallbacks

🔗 链接