dsh-micro-inversion-standard
3 ★
一个可复用的 DSH Agent Preset:双阶段、Token 精益的编码智能体模式。 核心目标:把模型思维链的起手习惯从 "let me" 翻转为 "we need",同时把上下文占用与 Token 消耗压到最低,并保持供应商 KV Cache 对未变前缀持续命中。
WorkspaceCoding
Author
DeepSeek Harness plugins published by @Shizuku-keop on GitHub.
一个可复用的 DSH Agent Preset:双阶段、Token 精益的编码智能体模式。 核心目标:把模型思维链的起手习惯从 "let me" 翻转为 "we need",同时把上下文占用与 Token 消耗压到最低,并保持供应商 KV Cache 对未变前缀持续命中。
DeepSeek Harness Brainstorming Visualization Plugin Seamlessly combining AI divergent reasoning with visual interaction, built specifically for solving complex problems
Compatibility governance for DeepSeek Harness: upgrade pre-flight gate, storage-format fingerprint, backup, session migration, per-profile lockfile, plugin x DSH compat matrix.
Session loop-health diagnostics for DeepSeek Harness: oscillation/stall/near-repeat/per-tool/token/compaction profiles + auditable 0-100 score. CLI + live watch bundle.