Reasoning layer for the official DeepSeek Harness: every request stays on native max effort, with isolated Flash/Vision/Pro depth controls, stability axes and reasoning toolboxes built above it. Delivered build only; commercial use open, reverse/modify/extract closed.
Extra model-request recovery for DeepSeek Harness: retries PI_AI_ERROR failures once after a fixed 20s delay; other error codes keep the built-in dsh-llm-retry exponential backoff.
Rewrites the User-Agent sent by every outgoing HTTP request (LLM API calls and other global-fetch traffic) to a value of your choice, configured live from a dedicated UA 设置 (User-Agent) tab in the Web settings. 为 dsh 发出的所有出站 HTTP 请求(LLM API 调用等走全局 fetch 的流量)改写 User-Agent,并可在 Web 设置的 UA 设置 选项卡中实时配置。
Processes tab for DeepSeek Harness: every agent and subagent of a conversation as a live console with paired tool requests and responses, installed as one profile bundle.
Solidify DSH session history step by step: rewrite tool results in place to the facts the model actually kept, so long conversations stop re-sending noise every turn. Costs tokens and time: a ~770-character contract in every request, plus one extra full-context request and round-trip per step when stepSummary is on.