DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
Agent skills for building and testing DeepSeek Harness plugins — from scaffolding a new plugin package to choosing the right test tiers, entirely inside an agent session.
Trusted Auto: a middle permission tier for DeepSeek Harness between workspace-write and danger-full-access, auto-approving harmless commands and trusted-area targets
DeepSeek Harness Web 侧边栏 DeepSeek API 额度卡片 + 充值面板插件 | Sidebar quota card & recharge panel for DeepSeek Harness: real-time official /user/balance (60s auto-refresh), recharge tiers/payment methods matching the official site, official top-up QR, API key stays in Host. One-command install: bash install.sh
Tiered auto-review for DeepSeek Harness: static-rule safety net + LLM reviewer + human fallback — auto-allow safe actions, deny irreversible ones, ask a human for the rest.
Every skill you already have — Claude Code, Codex, Cursor, Gemini CLI — works in DSH: Agent Skills SKILL.md discovery, Tier-2 conversions, find_skill search, and a skills doctor
A client plugin for the DeepSeek Harness Web GUI that adds chat content width tiers with a floating picker button (bottom-right, above the composer): 标准 / 中等 / 宽 / 超宽 / 全宽 (standard / medium / wide / ultra / full).
🌐 Multi-provider web search plugin for DeepSeek Harness: Zero-key Bing/Baidu direct search, free-tier APIs (Tavily/Brave/Serper/Bocha), SearXNG token auth, auto-fallback, and a built-in Web Settings GUI.
Home of dsh-plugin-hub: discover, evaluate, and install DeepSeek Harness (dsh) community plugins from inside dsh — plugin_search / plugin_info / plugin_install / plugin_remove tools, supply-chain trust tiers, and a Web Plugins marketplace tab. Also powers dsh.qomob.ai: a Chinese 0-to-1 Wiki and a daily-updated dsh plugin directory
Long tasks shouldn't cost like they need a frontier model. Cheap-model long-task engine: reliability in structure (disk state machine + mechanical gates + T3 zero-prose dispatch + multi-brain audit), not in model strength. Pure Python stdlib. DSH skill.
The auto permission tier for DeepSeek Harness: workspace-write requests run directly, sandbox escalations (danger-full-access) are judged by a separate classifier model.
Auto-approval permission guard for DeepSeek Harness: a middle tier between workspace-write and danger-full-access — auto-allows safe ops in trust dirs, always prompts for destructive ones
Automatic model routing for DeepSeek Harness: three difficulty tiers (hard/normal/easy) plus vision routing, picking models you already configured under Settings ? Models. ???? + ?????????
