플러그인

웹 UI에서 대화 파일 링크 미리보기(이미지/텍스트/Markdown/구문 강조 코드/git diff) — 링크를 클릭하면 데스크톱 오프너 대신 미리보기 모달이 열립니다. 루프백으로 격리된 라우트, 약한 ETag/304 캐싱, DOMPurify 새니타이즈.

Fail-closed 방식의 Feishu(Lark) 채널 브리지: 봇과 채팅하면 dsh 에이전트의 응답을 받습니다 — 옵트인 방식의 휴먼인더루프 도구 승인 포함(모든 bash 호출 전 Allow/Deny 카드, fail-closed 타임아웃). 공식 Python SDK만 사용하는 통합(버전 고정); 기본 거부 화이트리스트, 웹훅 서명/타임스탬프/재전송 검증, 채팅별 고정 세션; 매일 최신 SDK 호환성 카나리아 테스트; 이중 언어 문서.

반복 루프 기능: `/loop` 명령 + loop 도구 + 활동 상태 표시줄을 제공합니다.

모바일 원격 스위트: QR 오퍼가 있는 Host 페어링 플러그인, 루프백 Host Gateway, 암호화 Tunnel 및 동일 LAN WebRTC Direct, 그리고 이 리포지토리의 Android APK.

Analyze usage across models, providers, workspaces, and time ranges — including custom start/end dates over the full retained history — with live cache-hit visibility and separate model and provider breakdowns. Includes a 53-week GitHub-style heatmap, usage streaks, cache-efficiency trends, DeepSeek account balance, workspace aliases, CSV export, and a durable ledger that preserves successfully flushed usage after session deletion and rebuilds incrementally — on restart, unchanged sessions are applied from the ledger without re-reading their logs (a per-session log revision is the change signal), and only changed/new events are folded. Token accounting is explicit: input is fresh (cache read/write and reasoning are separate buckets) and zero-usage replays never overwrite recorded usage. Chinese uses local time, English uses UTC; API access is loopback-only.

HTTPS + login gateway for the DeepSeek Harness web surface: access your DSH Web UI securely from another machine on your LAN, or from the internet via port forwarding. Fail-closed defaults (loopback-only, no accounts until configured); setup and security notes in the README.

루프백 우선 방식의 멀티 에이전트 메시지 릴레이입니다: HMAC 인증 브로커와 dsh 플러그인(relay_send/recv/peers/history), 의존성 없는 CLI 및 Python 클라이언트, wire protocol v1.0을 제공합니다.

에이전트 트래젝토리 거버넌스 및 이상 진단 기능입니다: 평면적인 세션 로그를 다중 분기 트래젝토리 트리로 재구성하고, 루프 데드락, 유효하지 않은 재시도, 목표 이탈을 감지하며 비용 귀속과 함께 경고합니다. 공식 API를 통한 원클릭 중단 및 브레이크포인트 포크와 독립 GUI 탭을 제공합니다.

Deep research as an agent mode: one reviewed loop that plans, researches in adaptive parallel rounds driven by the gaps the researchers themselves declare, synthesises a cited report and then reviews it adversarially, instead of a script the model rewrites per call. Questions the round budget never reached are named in the report rather than dropped.

DeepSeek Harness용 로컬 우선 AI 프로젝트 오케스트레이션 워크벤치 및 CLI 플러그인: 승인 게이트 플래닝, Git worktree, 작업 실행, Issues 및 감사 가능한 증거.

Capability transform & assembly engine (Qian Xuesen cybernetics/systems framework): static resident layer manages the skill/MCP/plugin registries, auto-absorbs skills and tool metrics, and drives a negative-feedback evolution loop (scan Sick/Weak → mutate → select promote/reject → 30s rollback); includes a Web UI panel and a GitHub dsh-plugin store bridge.

Long-task orchestration loop: Codex plans/scores/reviews, DeepSeek executes, Kimi supplements (vision acceptance, summarization, cross-check).

Turns the iterate skill into an autonomous closed-loop harness — plan, parallel review, atomic fixes, validate, loop, auto-stop — plus a dry-run review-only mode and a meta-review that audits the report and emits a final review report.

Free voice closed loop for the Web UI: browser SpeechRecognition mic input with live interim results plus read-aloud speaker buttons and auto-read for assistant replies, zero configuration and no API key.

SearXNG-backed web_search provider for the web seam: free, self-hosted, key-less metasearch through your own instance's JSON API, with a bundled loopback-only docker-compose example.

파인만 학습 모드 플러그인: 가르치기 → 되짚어 설명하기 → 평가 → 재설명의 루프를 리치 HTML 수업 카드로 렌더링(mermaid 다이어그램 + shiki 코드 하이라이팅)

Markdown 노트를 위한 이중 지식 그래프 RAG: 선행 조건 + 유사도 그래프(P/S-Agent 순회), 단락 청킹, n-gram/임베딩 하이브리드 검색, 명시적 가중치 피드백 루프 — MCP를 통해 제공

AgentDuel is a code-battle game. This plugin lets you complete the entire game loop inside DSH: manage deathmatch characters and capture-the-flag teams, submit agent code, start battles, watch replays, and use DSH conversations to keep improving agents based on battle results.

DeepSeek Harness (dsh) plugin embedding the fin-trace agent loop — multi-hop relationship reasoning over a financial knowledge graph, exposed as async tools (start/status/cancel)

Keeps a local task-outcome ledger for DSH, verifies acceptance criteria from deterministic evidence, records user disposition and token usage, and supports digest-bound export.

Exposes dsh's internal capabilities (language, workspace registry, agent-status and approval events) over HTTP under /dsh-api on the local loopback socket dsh already listens on, with SSE fan-out at /dsh-api/events.

Auto-continue an interrupted dsh session after a web restart by injecting a "continue" follow-up; one-shot per process, no duplicate-injection loops.

Human-in-the-loop tools (`human_task` / `human_task_ready_check`): pause the agent until the user performs a real-world action or returns an observation (GUI operation, game testing, visual verification, hardware checks), gated by session consent and an AFK presence check, returning a structured JSON result via a web task dialog.

Play your own local music in the DSH web UI: the host scans a local folder (default ~/Music) and streams it over HTTP, with a now-playing bar (prev/play/stop, loop modes, volume, live spectrum) and a draggable playlist panel, plus a model-invokable `music_play` tool.