11 results
Weak-network adaptor for the DeepSeek Harness: transparent long-backoff model-stream retries (10-min cap), local response replay cache, heartbeat auto-reconnect, degraded-mode token economy, and a bilingual (zh/en) settings page. / 弱网适配插件：模型流长退避重试、本地响应缓存、心跳自动重连（默认 10 分钟封顶）、降级省 token、双语设置页。
Context compression tool (context_compact): the agent writes the replacement checkpoint itself and hands it to the host compaction engine, which skips the LLM summarizer call; automatic compaction stays on the official engine.
Cache-aware Harness Alliance Mode for DeepSeek Harness with DSH-routed Claude Code, Codex, and Kimi Code execution.
为 DSH read 工具提供版本令牌校验的进程内结果缓存：未变更文件的重复读取跳过文件 I/O 与重渲染，并附带 readcache 工具查看统计与清空缓存。
Prompt-cache observability and logical-input diagnostics for DeepSeek Harness
A DeepSeek Harness plugin: model-facing reload_plugin tool that restarts one Cordis Loader entry (by entry id, module name, or MCP serverName) without touching sibling entries — respawns mcp-client server processes and hard-reloads in-process plugin code (ESM/CJS cache bust + fiber swap)
Reasonix-style cache-aware compaction backend for DeepSeek Harness (DSH). Replaces/enhances compaction-basic with compact_ratio, one structured summary checkpoint, and a stable recent tail.
A tiny dsh web plugin that shows the cache-hit rate with two decimal places in the chat stats line.
把 N 次 LLM 调用压成 1 次 —— DeepSeek Harness（DSH）的极致调用压缩插件：标题短路 + 飞行去重 + 24h 回放缓存 + 批量归并。
Tiered micro-clear layer for DeepSeek Harness compaction: clears whitelisted re-runnable tool results to placeholders before the official summarization tier, freeing context with no model call.
Harness-native J-Space cognition preset with cache-stable scoped prompt control.