Plugins

Complete LAN/remote access for the Web UI: 0.0.0.0 binding, crypto.randomUUID polyfill, token gate (401 login page + WebSocket interception, loopback exempt), privileged-fence and settings-persistence exemptions, and an idempotent installer with upgrade recovery.

Exposes bounded Prometheus metrics for DSH sessions, agents, LLM calls, tools, approvals, subagents, and jobs, with a Grafana dashboard and a loopback-only endpoint by default.

Self-healing runtime layer: a resource control loop with strategy switching and repair escalation, error-storm and fiber-failure insight, cross-session knowledge carry-over, a self-healing dashboard and 7 tools.

One-click start/stop button pinned to the window bottom-right corner: restarts DeepSeek Harness in place via a detached helper that relaunches the exact boot invocation (loopback same-origin guarded), with an optional stop-only action.

Automatic type-check diagnostics: after the model edits or creates TS files, runs `tsc --noEmit` in the background (debounced per project) and reports errors via the `code_check` tool — grouped by file with line/column, severity and TS error codes, closing the "edit → check → fix" loop.

A project-agnostic multi-agent collaboration protocol for the DeepSeek Harness: loop guards that stop runaway agent loops, a six-field handoff format, risk-tiered review routing, and a fixed verify→commit→push→review→close sequence. One skill instead of the same rules copy-pasted into every agent's instructions.

Persistent wallpaper skin for the dsh web profile: static image or muted looping video as the app background, with translucent panels and a floating picker panel.

Incremental diff reviewer for DSH: after approving a batch, only new changes since the checkpoint are shown, with a Web review panel and `/review` command that inject feedback back to the agent.

A model-native meta-agent plugin for DeepSeek Harness: the model synthesizes task-specific workflows at runtime from five agentic workflow patterns (prompt chaining, parallel workers, router, supervisor, evaluation loop) and coordinates tools and subagent

Floating power dock for the DSH web GUI: one-click restart or graceful shutdown of the dsh host, loopback-guarded and dependency-free.

Per-profile settings overlay for DeepSeek Harness: global settings.yaml stays the baseline while each profile overrides any namespace through its own profiles/<name>/settings.patch.yml — object sections merge recursively, scalars and arrays replace wholesale, and !unset masks an inherited value. The overlay is transparent to existing plugins (they keep reading ctx.settings unchanged), writes land in the profile overlay only, and official schema validation, revision semantics, expectedRevision conflicts, watchers and events are untouched. Ships a settings command family (get/set/unset/mask/unmask/promote/demote/migrate/diff/layers) plus a Profile Settings section in the web Settings panel over a loopback RPC channel.

dsh web 客户端插件：settings.* 配置平面（describe/update/replace/mutate/openDocument）在非 loopback 页面也可读可写——修复经域名反向代理访问时，模型/提供方目录页报 “settings are unavailable in this browser” 与 “/api/settings.mutate: HTTP 403”

Dual-model auto-routing plugin: low-cost direct / high-cost upgrade with an escape-learning closed loop (wrong direct answers auto-learn fingerprints, force-upgrading the same fingerprint next time), persisted and interoperable with the Python ModelRouter.

Exposes the DSH dynamic-Cordis toolset to Claude Code over MCP (streamable HTTP on the existing web port, no new listener): ten tools to list, inspect, define, run, stop and permanently remove dynamic plugins in a live DSH session, with the same approval flow the GUI uses for packages that ship a client half. Bearer auth is mandatory — the token is configured on a DSH settings page, the endpoint answers 503 rather than serve without one, accepts loopback only and never emits CORS headers.

Three-tier agent memory harness: L1 always-on CLAUDE.md, L2 index, L3 detail skeleton with templates, a gate that blocks writes to protected files without a one-time approval token, zero-judgement event logging, weekly distillation, and a TDD execution-loop skill.

macOS desktop shell for the DSH web UI: a plugin plus a small Tauri/WKWebView window over the loopback web surface — real macOS fullscreen, single instance, external links delegated to the system browser, and browser-style zoom.

Durable batch review and commit safety for DeepSeek Harness tool side effects

Self-learning loop for dsh: record lessons, draft and promote skills, periodic curator review that corrects, consolidates, and retires skills.

DeepSeek Harness (dsh) approval answerer backed by the agent-governance sidecar — every approval/request is decided by the deny-by-default gate, L0–L4 rules and the organizational audit chain.

DSH Web Usage Center with local SQLite history, HUD and dashboard, Hub account/quota snapshots, custom pricing and fee ledger, and CSV/JSON export, plus coding-plan OAuth and an optional loopback OpenAI/Anthropic gateway.

Falsification-driven correction loop for agent sessions: risky work commits an evidence-checkable claim first, deterministic gates verify it, falsified claims force mutually exclusive replacement hypotheses with discriminating experiments, and every event lands in an append-only evidence ledger.

Guardian state engine for the DeepSeek Harness — assertion counting, edit review queue, and no-feedback-loop monitoring over session events

DSH plugin: password gate for remote (non-loopback) /api access (passwordless by default), unblock the loopback-pinned privileged endpoints, restore everything on uninstall, and give the local 127.0.0.1 user the native OS folder picker for workspace selec

Email reminders from your DSH agent when you are away, with stable SMTP (no 535 retry loop), connect and idle timeouts, and a --check self-test.