
DeepSeek Harness (DSH) ecosystem: curated plugins, tools, and infrastructure from dsh-external/hub a
DeepSeek Harness (DSH) ecosystem: curated plugins, tools, and infrastructure from dsh-external/hub a
English | 简体中文
Install Contribution guide DeepSeek Docs Public plugin topic Issues 完整目录
Curated DeepSeek Harness (DSH) ecosystem: plugins, tools & infrastructure. Sources: dsh-external/hub catalog and the public GitHub dsh-plugin topic.
Note: the GitHub
dsh-plugintopic is public; somedsh-externalrepository links may still require org access.
Install the official runtime with Node.js:
npx @deepseek-ai/dsh web
Install an external profile bundle with pnpm on your PATH:
dsh plugin --profile web add "github:owner/repo#ref"
dsh plugin forwards package operations to pnpm, so npm, Git/GitHub, local path, file: and link: package specs are supported. Only packages declaring dsh.bundle.patch become active profile layers; plain dependencies remain installed but inactive. Restart dsh --profile web after installing or updating a bundle.
The former &path: sub-path and Repository Plugin installation forms are not part of the current official bundle flow; use an installable package that declares dsh.bundle.patch.
Management panel: Settings → Plugins.
NanmiCoder/dsh-agent-teams - Multi-agent team orchestration: captain-style session with durable sub-agent teams, role assignment, and coordinated task execution.
dsh-todo-guard - Reliable todo panel that survives restarts (official panel only re-renders on write — fixed via projection pre-warm) with three-state completion verification: evidence exists → verified, fake evidence → blocked, no evidence → unverified badge; settings toggle to fall back to official behavior.
fakechris/dsh-track - Embedded task-management engine for DSH: decision-point protocol, idea capture wall, Linear-shaped issue store with an evidence-driven lifecycle.
dsh-dual-model-eval - Runs one coding prompt across multiple configured models in isolated Git worktrees, streams side-by-side tool traces and results, and commits the candidate the user adopts for later rounds.
dsh-agent-arena - Compares coding agents in isolated Git worktrees with deterministic validation, scoring, and explicit winner application.
xiehuan123/coding-coach - Coding Coach: 35-skill bundle plus a full agent preset for non-developers (8-stage idea-to-launch pipeline; engineering/product/UI skills).
dsh-collaboration - Multi-agent collaboration suite: user-configured specialist roster, persistent on-demand dispatch (team_call/team_message/team_status/team_close), clone instances, star-topology relay, model comparison and a multimodal vision bridge.
dsh-plans - Planning-first agent preset: research repository changes into traceable Markdown plans, refine them through reviewer/criticizer subagent rounds, then execute as a DSH goal with a verifier checklist.
dsh-agent-team-gui - Persistent multi-model squads managed in Settings and selected in the Composer; the lead Agent dynamically plans bounded DAG runs with optional review/repair, while Run Center reports DSH's official provider-reported Token usage.
DSH Automation Center - Workspace automation center for scheduled Agent runs: each execution starts a fresh Result Session with persistent audit history; stock DSH uses a Conversation tab, while compatible Shell slots enable a global page.
Knotline - Visual DSH project map for composing persistent agent workflows from requests, agents, skills, backlogs, approval pools, and scheduled triggers.
cleverer-dsh - Execution-discipline suite for DSH with identical-retry interception, forced reflection, todo enforcement, memory deduplication, and experience-to-skill promotion (11 plugins + 6 skills).
february2015/dsh-taskswarm - DSH port of TaskPlane: dependency-ordered waves run in parallel git-worktree lanes, with task packets, cross-model review, and crash recovery.
hongyue0721/dsh-kimicode-swarm - Kimi Code Swarm-style batch parallel subagent dispatch: a swarm_batch tool fans independent subtasks out to real subagents with two-stage adaptive concurrency (ramp-up plus collision-driven exponential backoff, auto shrink/recover), an in-chat live progress stream (SSE), and resume_agent_ids continuation.
timwhitez/dsh-self-evolving - Evidence-first, crash-resumable self-evolution engine for DSH: generates bounded Cordis plugin candidates, admits them through a one-shot real Loader, evaluates with Harbor, and journals an auditable lineage.
Saktawdi/dsh-ha-orchestrator - Model high-availability failover (quarantine, circuit breaking, probe recovery) plus subagent orchestration (fanout/pipeline/supervisor) with a bilingual settings UI.
dsh-background-agents - Durable background child agents on the official subagent seam: start from any session, watch progress in the Web UI sidebar, message and interrupt any time, with per-child tool scoping, persona and delegation-depth caps.
dsh-team-rooms - Persistent, cross-session multi-agent team rooms for DeepSeek Harness: members, a message bus, a shared task board, approval-gated handoffs, and a shared timeline that survive restarts.
zoahdev/dsh-kirocrew - Bridge a DSH agent to a persistent, self-evolving KiroCrew development workspace over ACP (JSON-RPC 2.0 over stdio) via a single kiro_send tool.
bpc-oss/dsh-routed-subagent - Run a one-shot subagent fully mounted on any agent preset from any session, with per-call model/provider override, model pre-check, and external CLI engines (codex / claude / codebuddy) with background jobs, live progress, kill, and continuable sessions.
bpc-oss/dsh-fork-to-preset - Fork any session into a different agent preset from the conversation header: a preset picker creates a new child session mounted on the chosen preset, inheriting the source session's completed turns.
qwert702/dsh-commander - Commander mode for DSH Web: inject protocol briefs into the session title bar, parse task blocks from model replies and auto-execute them, separating strategy from execution; activated via a badge button.
dsh-product-subagent-console - Conversation-level multi-agent workbench for editable task planning, real child-session observation, plan-versus-runtime comparison, and evidence-backed recovery previews; tested with DSH 0.1.1-rc.2.
weibaohui/dsh-continue - Auto-resume for interrupted agent sessions: an ordered rule table routes by failure type (rate limit, quota, auth, context overflow, crashed orphan) into backoff retry, model switch, resume after compaction, or stop-loss notification, with a visual rule editor and activity log.
weibaohui/experts-management - Expert manager: manage ntd-format experts and expert teams (plugin.json + A
No open issues yet, or sync has not completed.