Orbit — a reflection layer for DSH: hypothesis→task→result rings with objective done_when, rolling review, track-scoped playbooks, and weekly distillation with review gate.
Agent 連携
$ npx -p @deepseek-ai/dsh dsh plugin --profile web add github:clawdwan/orbit-dsh-plugin
Multi-agent orchestration for DeepSeek Harness: delegate, adversarial review, diverse explore, background runs + converge. One prompt. A team of agents.
スキル · Agent 連携
$ npx -p @deepseek-ai/dsh dsh plugin --profile web add github:ConradLu2740/dsh-orchestrate
DeepSeek Harness plugin: a native-looking modal on task send lets you pick a different model for subagents and an independent reviewer (own model) that validates each subagent's work, fixes only real defects, and delivers the result to the main agent.
Agent 連携 · 開発ツール
$ npx -p @deepseek-ai/dsh dsh plugin --profile web add github:frederico-kluser/dsh-orquestrator
DeepSeek Harness agent presets whose leader does no work itself: every non-management tool call is denied at the harness boundary and the task goes to a subagent instead, after a one-time research pass over the available subagent models and a per-task model choice (a second preset deliberately picks the model least suited to the task).
Agent 連携
$ npx -p @deepseek-ai/dsh dsh plugin --profile web add github:Funny1Potato/dsh-outsourcing-expert
Embedded Overleaf workbench tab for DeepSeek Harness Web: same-origin reverse proxy for self-hosted or public Overleaf (no X-Frame-Options limits), direct-CDP login with credential storage, selection quoting into the composer, agent-driven cursor insertion, and LaTeX workflow helpers
Agent 連携 · 生産性ツール
$ npx -p @deepseek-ai/dsh dsh plugin --profile web add github:gychen-NJU/dsh-overleaf
An eye that grows: accumulated design taste for DSH agents. A taste corpus + codified principles that sharpen with every example you feed — review designs against learned judgment, not a fixed ruler.
Agent 連携
$ npx -p @deepseek-ai/dsh dsh plugin --profile web add github:guo6x/dsh-palate
Password generator client plugin for the DeepSeek Harness Web GUI: random passwords (low/mid/high tiers) plus an Agent-generated meaningful tier. No history is persisted anywhere.