플러그인

DeepSeek Harness 에이전트 기능을 MCP 서버로 노출해 외부 MCP 클라이언트(예: Hermes)가 Harness를 구동해 코딩 작업을 수행하도록 함. Hermes = 두뇌, Harness = 팔

Codex, Claude Code, WorkBuddy, CodeBuddy 메타데이터에 대한 제한된 프라이버시 안전 로컬 스캐너: 키 기반 증거를 원자적으로 집계하고 내장 DSH 프리셋 및 선택적 기능을 결정론적으로 권장 — 콘텐츠 보존, 네트워크 액세스 또는 프리셋 변경 없음.

Graph runtime capabilities for the DeepSeek Harness: register LangGraph compiled graphs as DSH tools, plus a graph-driven routing agent.

Nearly half the agent cost, with accuracy intact. Just enough tools is a DeepSeek Harness plugin that uses Jev to reveal tools and skills progressively. The main model starts with a clean planning step. Jev then selects which capabilities to add as the task unfolds.

Fronts stdio MCP servers with three router tools (list_capabilities, describe_tool, call_tool) so schemas load on demand instead of riding every request; shapes every tool result through a deterministic output ladder that archives the original first (retrieve_spill reads it back); prices tool groups in tokens and can hide built-in groups per agent. Bridges to a standard-library Python router, so Python 3.9+ is required.

Lets the AI insert streaming-generated UI into the conversation or a side panel, written in TSX — already one of the model's native languages, so there is no JSON schema to teach it. Rendered without flicker, state kept across frames with no hard remount, and a click in the card driving the next chat turn. Inline is where streaming matters most; the side Canvas is the more agentic shape. Cards get agentic capabilities of their own — run bash, curl, call the model, read and write files.

로컬 'Screen Automation Helper' 데스크톱 플랫폼(Windows/macOS)을 에이전트 도구 시스템에 연결하는 DeepSeek Harness(dsh) 플러그인. 상태, 기능, 워크플로 목록/실행/중지, 실행 관리, 화면 캡처, 로컬 인식 프리미티브를 노출합니다.

dsh의 내부 기능(언어, 워크스페이스 레지스트리, 에이전트 상태 및 승인 이벤트)을 dsh가 이미 수신 대기 중인 로컬 루프백 소켓의 /dsh-api 경로 아래에서 HTTP로 노출하며, /dsh-api/events에서 SSE 팬아웃을 제공.

OpenAI 호환 API 호스트에서 모델 목록, 가격, 기능을 자동으로 발견해 정규화하고, dsh용으로 바로 쓸 수 있는 모델 구성을 출력합니다.

Audits a DSH profile for capabilities nothing consumes: inventories the declared composition rows, counts how often each registered tool and skill is actually called in the session logs, separates calls that were made from calls that returned a result, and reports the ones with no observed consumer. Ships a skill that fixes the evidence-chain format for completion claims.

Computer Use for macOS desktop applications: reads any app's accessibility tree as text with element indices, then clicks, types, pastes, scrolls, drags, selects text and sets values through it, plus screenshots read back as images. Driven by a persistent JavaScript REPL — the agent gets two tools (`js`, `js_reset`) and calls an `sky` API inside them, so state survives between calls and the tool schema stays fixed as capabilities grow. Accessibility trees come back as diffs after the first read; actions address elements by index and indices from a stale read are refused rather than misapplied; typing waits for keyboard focus and reads the field back; paste restores the clipboard. Requires macOS 14.4+, Accessibility permission (Screen Recording only for screenshots) and Node 20+; ships a prebuilt native module, so nothing compiles at install time.

Runtime observability and plugin operations panel for DeepSeek Harness: shows the profile plugin tree, real Loader fiber state, capabilities/dependencies/compatibility, current-session plugin activity and upstream source/version checks, with one-click enable/disable/update/uninstall plus a standalone startup-failure diagnostics page that works while DSH is not running.

Native DSH foundation bundle that registers a persistent default workspace, provides paged read-only inspection of plugins, skills, memory, sessions, storage, settings, credentials, and DSH_HOME files, and synchronizes rich CLIProxyAPI model capabilities into native provider settings.

一个DSH插件，为自定义模型配置输入类型(input)、推理模式（reasoningEfforts）

DSH plugin: auto-detect and configure model capabilities (reasoningEfforts + input modalities) for llm-pi-ai. Successor to dsh-reasoning-efforts.

Per-model reasoning effort (off/minimal/low/medium/high/xhigh) for GPT, Codex, Gemini and OpenAI-compatible models. Installs a helper, not a runtime adapter: run enable-capabilities.ps1 once per profile (Windows) and restart — it declares the levels in ~/.dsh/settings.yaml, with a disable script and a backup to revert.

LLM compatibility auto-healer: detects 400 errors from incorrectly declared model capabilities (thinking/reasoning) on transit/gateway providers, then idempotently patches `llm-pi-ai` config and suppresses thinking on the next request — no restart needed. Full settings UI for every pi-ai compat field.

Hold a DeepSeek Harness agent to a capmark capability manifest: mask its tools and judge every call.

Configure model input capabilities and reasoning efforts for DeepSeek Harness

Container-isolated backend for the `ctx.codeRuntime` seam: each Code Mode program runs in a fresh container with no network, a read-only rootfs, dropped capabilities, and kernel-enforced memory, CPU and pid ceilings.