Plugins

Human-in-the-loop self-improvement loop: one objective becomes an auditable work chain that iterates in child sessions while the main agent stays idle.

Experiment-orchestration SOP skill for long-running experiments. A top-level model drafts a decomposition plan — subtasks, the model for each, an acceptance figure with tolerance, and dependency order — and dispatches only after the user confirms that plan; each subtask then goes to a lightweight executor subagent as a self-contained prompt, and its report is confined to measured facts and the delta from the agreed figure, with judgement and decisions staying at the top level. Ships one plan template and five task templates (gather, execute, consistency check, read-only review, mechanical transform), acceptance gates split into a task layer and an experiment layer, and three rules that stop internal-consistency evidence from being presented as independent or external validation.

Agent skills for the Lark/Feishu CLI: exports the 27 skills embedded in the lark-cli binary (docs, sheets, base, calendar, drive, task, wiki, approval, mail, slides, vc, minutes and more) and registers them with the DeepSeek Harness skill registry so the agent can read them on demand. Excludes lark-im so it complements an IM bridge plugin rather than duplicating one. Requires lark-cli to be installed.

DeepSeek Harness port of the Impeccable frontend-design skill: 24 design playbooks covering shaping, critique, audit, polish, layout, typography, motion, accessibility and design-system documentation, driven by the reference files shipped in the skill plus PRODUCT.md / DESIGN.md in the project instead of a downloaded engine binary. Registers the impeccable skill; reads project files and writes design artifacts under .impeccable/. The binary-only features (live browser bridge, pixel measurement, font matching) fall back to static variants, screenshots and explicitly declared manual estimates. Ported from pbakaus/impeccable (Apache-2.0).

E-commerce order analysis skill for DSH: refund-type classification, return-rate statistics, interception and refund-abuse loss accounting, after-sales analysis with product warnings, delivered as a single Excel workbook with a per-order detail sheet and a sampling verification sheet (the agent builds it with Python: pandas, openpyxl, matplotlib).

Provides a PMNote skill for finding public resources on game project management, careers, and AI tools.

Installs a TeleAgent skill-authoring standard as a DSH skill: the full spec, a validator that reproduces the platform checks (and fails on .pyc bytecode), a scaffolder, a packer and a TeleAgent installer.

Guide an agent through exporting selected OneLap cycling FIT files, validating data and coordinates, and uploading via a logged-in Strava browser session. Needs Python 3.9+ (standard library only) and an agent with browser and file-upload tools.

Matches each new task against installed skills with deterministic keyword scoring and injects the best one without being asked; records effective operations as per-skill notes for later runs.

Local Markdown notes panel: import any local folder, browse its Markdown tree in the sidebar, edit files with auto-save, and search across every imported folder. Shows read-only Git context for each file — status, diff against HEAD with per-line change bars, and line blame — and exposes six notes_* tools so the agent works on the same folders. Files stay where they are on disk; the plugin never commits, stages or checks out.

Runs loudness normalization, silence trimming, format conversion and batch processing through local ffmpeg, with presets for game sound effects, music uploads and voice cleanup.

Grabs a live frame from the local webcam and hands it to the agent as an image, so the model can actually look through the camera without opening any app or keeping a window in front. A dim room no longer comes back black — brightness, gamma, contrast, sharpen and an auto-level pass run inside the capture itself. ffmpeg is fetched automatically on first use, and the same plugin lists the available capture devices.

Drives the Edge already open on your machine through a local WebSocket bridge and an unpacked MV3 extension: the agent reads the page as Markdown, dumps its interactive elements with stable refs, clicks and types with pre-action assertions and a three-valued effect state, and can capture pages from the extension popup as self-contained HTML snapshots, filed into a kb.py literature knowledge base when one is configured (kbRoot or DSH_KB_ROOT). Seven browser tools stay unregistered until edge_bridge is started, so an idle session pays no context cost.

DeepSeek web search with its own API key: a left-sidebar search page and a deepseek_search tool that every model in the profile can call, falling back to the DeepSeek key already configured for the model provider. It registers its own tool and routes instead of replacing the built-in web_search and web_fetch providers, so an existing search setup keeps working.

Let agents read and control the web page open in the right Sidebar Browser tab — read text, run JS, click (HTML and SVG), type, scroll, screenshot, with a per-command token bridge behind the Harness connection fence. Full control and screenshots need the DSH Desktop app; in a regular browser (Web) only same-origin pages can be read or scripted.

Keeps the Mac awake for as long as the DSH host is running, whether or not an agent is working: starts ClamAwake's virtual display (mirroring the main screen) plus caffeinate -d, stops both when DSH exits and restarts the engine if it crashes. Self-contained (bundles the arm64 vd-engine, GPL-3.0). Requires macOS on Apple Silicon.

Serial model orchestration: a delegate tool over a built-in catalog of current Chinese-market models (DeepSeek, GLM, Doubao, Hunyuan, MiniMax, MiMo, Kimi, Qwen, GPT). The main agent splits work, picks a model per subtask with a required reason, waits for each result, and falls back to a default model; every route is checked against the Session model-selection policy before a child starts. Needs Plugins → Subagent → Model selection enabled for the routes you want (new sessions only) and each model's provider configured. Uses the same plugin id as other model-router plugins, so install only one of them.

Connects the Tabby serial MCP bridge and registers a tabby-serial skill: the agent can list, open, read and write the serial ports already open in Tabby, and the same terminal keeps accepting your keystrokes while it works. Requires the tabby-serial-mcp plugin inside Tabby.

Runs a shell command on your machine from the chat box with !cmd or /sh cmd, without sending a model request. The output comes back as your message plus a terminal card with the exit code, run time and a Copy button. While a ! line is typed, a menu offers zsh completions from your own fpath and matching commands from your shell history. Commands run as a login shell in the session folder, outside the agent sandbox, with a 120 s default timeout.

Standing GitHub-first open-source check: a system-prompt section makes the agent extract 1-4 keywords on every new request and make one cached, rate-limit-aware github_search call, so the answer can list up to five reusable repositories with stars, language, license, last push and topics.

Local-first AI literature desk for DSH: import and parse PDFs (MinerU APIs or on-device pdf.js), read original/bilingual text with in-margin annotations, ask grounded questions whose citations jump to the page, and build per-paper or library knowledge graphs; eight agent tools expose the same library.

Model-classified tool compass for DeepSeek Harness: the agent sees functional categories and pulls specific tools on demand.

Daily check-in for MiYoushe (米游社) across six miHoYo games (mainland China servers only): Genshin Impact, Honkai: Star Rail, Zenless Zone Zero, Honkai Impact 3rd, Tears of Themis and Honkai Gakuen 2. QR-code login inside the plugin panel, per-game selection, a daily schedule with random jitter and startup catch-up, and four agent tools. Credentials stay on the local machine (mode 0600) and are never sent anywhere but the official miHoYo API.

Adds one agent tool (read-only on your transcripts) that searches Claude Code, Codex and DSH session transcripts and returns each match with its resume command. Needs ripgrep (rg) and zstd (zstdcat) on PATH and keeps a plain-text cache of DSH sessions under ~/.dsh/cache/history-search.