플러그인

DSH를 Timeline Studio에 연결하여 `.timeline` 프로젝트 검사, 의미론적 편집 미리보기, 트랜잭션 편집 및 검증된 MP4 렌더링을 지원합니다.

모드별 프롬프트(caption, UI, 문서, grounding, topology 등)를 통해 대화 이미지를 분석하고, 좌표 프리미티브(박스, 포인트, 참조)와 함께 구조화된 증거를 텍스트로 주입하며, 세션 수준 캐싱을 통해 replay 및 compaction을 거쳐 재사용할 수 있습니다.

DSH용 로컬 개발 서비스 콘솔: 머신에서 수신 대기 중인 모든 서비스를 나열하고, 개발 중 시작된 서비스를 검사, 안전하게 중지하거나 재시작할 수 있게 합니다.

harmonyos-dev-mcp Python MCP 서버를 DSH에 연결: 디바이스 검색, 빌드/설치/실행/제거, UI 자동화(클릭/입력/키/스와이프/스크린샷), E2E UI 트리 검사, 로그/크래시 분석을 18개의 mcp__harmonyos__* 도구로 제공합니다.

모든 DSH 확장 지점을 하나의 선언적 DSL 뒤로 통합합니다. 런타임 `fabric` 서비스, `fabric_extend`/`fabric_inspect` 도구 및 역량 그래프 설정 페이지를 제공합니다.

dsh용 spec-driven 개발 규율: 5개의 keel 스킬(anchor/spec/probe/build/audit), 3개의 리뷰 도구, 6개의 spec 템플릿이 코드보다 먼저 명세를 쓰는 것, 가정 검증, 범위 통제를 강제합니다.

에이전트 산출물을 위한 독립 브라우저 승인 테스트입니다. JSON 사양을 입력하면 실제 Chromium 판정이 출력됩니다.

Unreal Engine .uasset을 검사(resolve/registry/load)하고, 내장 UE Python으로 제한된 Blueprint 노드 편집을 수행합니다.

DeepSeek Harness용 모델 네이티브 메타 에이전트 플러그인: 다섯 가지 에이전틱 워크플로 패턴(프롬프트 체이닝, 병렬 워커, 라우터, 슈퍼바이저, 평가 루프)으로부터 모델이 런타임에 작업별 워크플로를 합성하고 도구와 서브에이전트를 조율

Cross-session memory as plain Markdown per workspace: weighted injection, bi-temporal invalidation, TTL review with stale markers, credential redaction, error and session retrospectives, SOP skills, and approval-gated self-improvement proposals.

Let a DSH model inspect and operate Windows applications through an installed Codex Computer Use runtime, with per-session app approval.

Visual sketchboard for DeepSeek Harness: draw with Excalidraw, let AI agents inspect and annotate sketches, and review proposed edits.

Read Windows service state from WSL: list services, or inspect one by name with its dependencies, logon account and binary path.

Read-only Skill inspector for DeepSeek Harness. It shows which Skills a conversation actually loaded — from observed loading evidence only — and which Skills the current environment can discover, and gives each Skill a detail page that keeps its declaration apart from its run: the body split into structure, declared workflow and progressive disclosure, the run logic read from local receipts, and the per-step evidence behind each claim. Evidence is never promoted into a conclusion — a step with no observed evidence is reported as such, not as executed or not executed — and the run's instruction hash is shown beside the current definition as match, mismatch or unavailable. It also reads the body, offers a Chinese reading version, and can clone a Skill into a local copy. It never writes back to the source Skill and never runs its scripts.

Auto-monitor and reveal active file operations and changes in DeepSeek Harness (DSH) right sidebar

Cordis-native, evidence-driven full-stack engineering workflow: inspect-first planning, explicit verification evidence, and approval-aware gating for sensitive operations.

Task-to-skill pairing meta-skill with a laziness ladder for minimal loading, a user-editable routing table, and a quarantine, SkillSpector scan, human-approval install flow.

DSH settings section for managing a local Sirchmunk knowledge backend: graph inspection, cluster management, audit history, runtime metrics, and privacy controls.

One MCP tool instead of every schema you own — a lazy capability broker that cuts resident tool context by a measured 94.4%. Search, inspect, enable and call MCP servers and skills on demand.

Browser patrol & website inspection plugin for DeepSeek Harness (DSH). Teach once, replay deterministic runbooks with managed Chromium, screenshots, credentials, checkpoints and selector recovery.

Coding retrospect distilled memory for DSH: distills rejected solutions, pitfalls and constraints from each turn via a local LLM, tags them into MemOS (type:rejected_solution), and boosts weighted recall on coding tasks via agent/pre-step.

WebUI panel for the dsh-guardian watchdog: start, stop and inspect two macOS LaunchAgents with status, log tail and explicit failure states.

Image generation workbench and agent tool for DSH: one tool (ark9_generate_image) paints text-to-image and image-to-image via any OpenAI-compatible API. Channels are entirely user-configured - no bundled endpoint, you bring your own baseURL, key and model name (manual input supported). Agent-initiated generations are approval-gated by default: requests queue in the floating panel and nothing bills until you approve; approval is non-blocking (the agent replies "awaiting your approval" immediately, and results, denials or timeouts are injected back so the agent reports them in the next turn). Floating glass workbench with five tabs (Generate / Approvals / Prompts / History / About), aspect-ratio grid with quality-budget + 16px-alignment sizing, transparent background, batch up to 10 images as aggregated sub-tasks, prompt library with custom JSON sources fetched through a host proxy, persistent generation logs with one-click retry, multi-channel aggregation with per-channel model fetching, bilingual UI (zh/en), stroke-SVG icon set, config import/export. Registers as a Better Sidebar tab when dsh-better-sidebar is installed; stacks above the dsh-cua FAB when both are present.

Three independent language channels for model output — tool-call descriptions (every tool-call card title, including run_code in PTC-style presets), thinking, and replies — each set to follow the web GUI language, force a specific language, or turn off (default: only descriptions, following the GUI). One global runtime-context prompt directive on the host; the client half reports the GUI language; a Settings plugin card switches the modes.