Plugins

Agent Context Editor is a cross-agent plugin for manually excluding, editing, and selectively compressing AI conversation context. It also supports searching, filtering, selecting, hiding, restoring, and undoing conversation changes while preserving the original Session.

Use the dsh web UI on your dev machine from an Android or iOS phone, connected peer-to-peer across networks via iroh; no server to set up, with a relay fallback that only carries ciphertext.

为 DeepSeek Harness 带来「跨会话长期记忆 + 后台自我进化」能力的纯插件实现：五轨记忆 · git 分支感知 · 回合内自我审查 · 技能自我进化与技能管理器 · 四轨待办 · COI 调度 · 会话广播 · 会话搜索 · 提示词管理器 · 临时信息便签——零核心修改、零运行时依赖，随装随用、卸载即净。

Catppuccin themes plus a toggleable glassmorphism skin for the DSH Web GUI: Latte, Frappé, Macchiato and Mocha registered into the native theme runtime, one-click switching with a persisted choice, and adjustable frosted glass for the top bar, sidebar, composer, stats line and trajectory view.

ChatGPT account sign-in for DeepSeek Harness through a Codex browser OAuth flow with no API key and no workspace selection: subscribes the login to pi-ai's Codex provider, lets you set the global default Codex model and reasoning effort for new sessions, refreshes the model list from the signed-in account's own catalog endpoint, and shows the five-hour and weekly usage windows in the composer.

Rewind a DeepSeek Harness conversation and its workspace together, without touching Git: every rewind is a fork, /redo reverses it, and a Rust engine snapshots in milliseconds.

Per-model reasoning-effort and input-modality configuration for third-party providers: suggestions fused from a model knowledge base and raw /models endpoint evidence with confidence labels, edited inside the official Models page card.

Claude Science-style research workbench: ReAct research-loop engine (research_* tools), versioned artifacts with provenance (artifact_* tools), and 10 science skills for genomics/pathogens/bioinformatics.

Mobile remote-control suite: bundle plugin with native sidebar entry and admin drawer, self-healing bearer-token gateway over LAN/Tailscale, Android app (sessions/approvals/questions/goals), /fs/* file transfer (Range resume, 2GB upload), latency-based multi-server auto-switch, offline chat cache.

Ports all 38 opencode themes (33 TUI + 4 desktop 2.0 + system) into DeepSeek Harness: data-driven pipeline (JSON → color resolve → DSH override), per-browser persistence, monospace/proportional choice, five code fonts, and a system fallback that keeps typography only.

Pause and resume tasks automatically during user-defined time windows to avoid peak-hour API costs; includes account balance display and 10-minute granularity spend bar chart.

Delete DSH sessions from the web UI and desktop client: header danger button and session-row menu item with a risk-consent dialog; host endpoint and agent tool remove the session log, projection cache, and workspace accounting.

Voice input for the web UI: tap Alt (or Alt+Space) to start/stop dictation, browser Web Speech (zero-config) or OpenAI-compatible cloud ASR, optional polish through DSH-configured LLM, settings UI.

Installs four multi-agent math-research presets for DSH: v2 (probability-driven pipeline with multi-verifier debate), v3 (paper-style Markdown knowledge base with a planner agent and a reusable method library), v4 (persistent self-organizing residents that message and meet), and v5 (a research institute with an academician who decomposes and assigns work, voting researchers, temp workers, group chat and a compare-and-set task board); all four support checkpoint resume, human intervention, and an optional Lean formal-verification switch (off/encourage/require) whose passing proof turns the vote into a fidelity check of the Lean statements.

Local-only project memory with zero bundled runtime dependencies or external services: JSON as the source of truth plus generated PROJECT_MEMORY.md, token-bounded session-frozen Hot Memory for prompt-prefix cache stability, cached cross-project BM25 recall, trusted session/turn provenance, explainable pre-write duplicate/conflict governance, editable importance/pin/tags/supersede/archive lifecycle, configurable top-level turn-end distillation, and a bilingual Web GUI for search, inspection, diagnostics, and live settings.

Pause and resume tasks automatically during user-defined time windows to avoid peak-hour API costs; includes account balance display and 10-minute granularity spend bar chart.

Generates game art from reference images via Seedream image generation and MiniMax image-to-video, assembling eight-direction sprites, sequence frames, and skeleton animations with local green-screen keying.

A draggable Live2D desktop pet for the DSH web GUI that tracks the pointer, changes motion, expression and outfit with the session phase, and opens a right-click control panel.

Input experience enhancement: voice input, AI polish, anti-overwrite, prompt optimization with diff preview, and auto locale switching following DSH UI language.

Text-first computer use for DSH: background Chromium control via Playwright/CDP plus accessibility-first macOS control; actions stay pinned to the right process and window without taking the user's pointer, ships a Developer ID signed, notarized Universal 2 DMG.

AI4Scholar academic search: 38 tools over Semantic Scholar, PubMed, Google Scholar, arXiv, bioRxiv/medRxiv, and DOI resolution (search, citation graph, authors, recommendations, PDF full text in slices, auto-cite, scientific figures), a unified cross-platform search with de-duplication, an API-key card in Settings, and per-call credit accounting with an `/ai4scholar` balance card.

Workspace-safe Office tools for agents: create/read Word, create/read/update Excel, and create/read PowerPoint decks with PNG/JPG/GIF image placement.

A Live2D whale-girl companion for the DeepSeek Harness Web UI: the model stays on screen and reacts to what the agent is actually doing (thinking, tool calls, streaming text, errors, turn completion); click her to send the agent a message and watch the reply stream into a speech bubble; right-click for a menu of expressions, dress-up items, desk scenes and one-shot animations, every action mapped from the original model author hotkey sheet so nothing overlaps and everything expires on its own. Non-commercial model assets (CC BY-NC-SA 4.0), MIT code, by Andersen216 (https://github.com/Andersen216).

Classifies pending tool calls with Jev (TypeSafe AI's non-generative decision model) on tools/pre-execute — confident high-risk calls are denied, ambiguous ones escalated to approval — and auto-approves clearly-granted reversible calls on approval/request behind argument-evidence gating. Also subclasses the session-reference resolver so snapshots keep Jev-scored messages instead of dropping oldest-first. Degrades to stock behavior on any provider failure; shadow mode with a /jev-stats command; TypeSafe or OpenRouter endpoints; 64 tests.