Plugins

Prompt enhancement for the DSH Web GUI composer: a sparkle button next to the send action rewrites the current draft into a structured prompt through the harness model, streaming the result into an inline preview (original vs enhanced), then fills it back with one click, keeps an undo bar, and offers a settings page for model route, temperature, length limits, system prompt, context awareness, undo-bar behavior and shortcuts.

Adds one button beside the composer that rewrites the draft into a more complete prompt in place, with no preview step, plus undo, a stop action, and optional conversation context.

Harness² recursive self-improvement for DSH agents: Propose–Probe–Compose, isolated probes, anonymous selection, and a per-domain experience library.

Give your DeepSeek Harness agent an addressable visual memory: revisit any image, inspect any region, and read exact pixels.

Route implementation work from the main model to executor models and collect their reports.

Chinese names and one-line explanations for installed DSH plugins; automatic translation and in-place display on the official Plugins page.

Multi-role group chat panel for DeepSeek Harness: join as one character, chat with 2-3 AI members, route replies by addressee with probabilistic bystanders, use @ handoffs, and keep private memory/relationships, a global character library, failure diagnostics, retry of failed replies, paged long-history rendering, cross-tab sync, provider-catalog models, and JSON import/export.

A transparent always-on-top desktop pet for DSH on Linux (X11 / XWayland), driven by the session state. The pet lives outside the browser as its own window, follows the pointer with a distance and per-display limit, shows a live session-status bubble that lists each session task and state, and auto-starts with your session. Linux only (X11/XWayland); needs Python 3 with PyGObject, GTK3, GdkPixbuf and xdotool. Its sprite sheets come from @michengai/dsh-codex-pet (MichengAI/dsh-codex-pet in this list).

Detects codex or Claude Code on the machine and exits the DSH host with code 66. Warning: on machines with either tool, installing it alone causes a boot crash loop; pair it with dsh-chaos-crash-fixer or set DSH_CHAOS_MODE=log. Joke plugin, not affiliated with OpenAI or Anthropic.

Exits the DSH host with code 1 when the dsh-chaos-crash-fixer repair plugin is installed. Warning: it crashes DSH on boot while the repair plugin is present; install dsh-chaos-fixer-crash-fixer to disarm it. Joke plugin.

Repairs dsh-chaos-fixer-crasher by disarming its exit trigger before it fires, keeping the four-plugin chaos chain stable. Warning: it only disarms this specific crash, not arbitrary uncaught exits. Joke plugin.

Repairs dsh-chaos-rivals-crasher by disarming its exit trigger before it fires, so DSH survives when codex or Claude Code is detected. Warning: install it together with the crasher; it cannot revive a host that has already exited. Joke plugin.

Detect and redact sensitive message text before sending it to models, with local regex, browser BERT, or ZeroClave API detection and manual review.

Hot reload and one-click restart for DSH plugin development. Reload uses DSH's built-in reload instead of a private reimplementation, so it behaves the way the host expects: only changed plugins refresh, the rest keep running. Restart is the safe one - it runs outside the client, so it still brings the client back when the client is what changed; it refuses while a session or job runs; it confirms the client reopened. Lightweight: one tool, no arguments, nothing added to the model's context, buttons off by default.

Experimental Windows bridge for Codex to supervise DSH Desktop 0.2.1-rc.3 tasks through local MCP, with blocking direction questions, plan checkpoints, review and revision, and named workspace routing. A Windows adaptation of Cryocell77/dsh-desktop-mcp-bridge (MIT); it uses the same package name and loader row, so it cannot be enabled alongside upstream.

A local task board with secretary, CEO, and historian roles, plus a desktop whale-girl companion for DeepSeek Harness.

Adds a terminology-aware correction pass to the built-in local speech-to-text: keeps DSH's own microphone and SenseVoice engine (audio never leaves the machine) and fixes homophones, domain terms and malformed numbers in the finished transcript.

Reads each reply aloud after rewriting it into spoken style: system TTS, Doubao speech synthesis, or an optional local Kokoro model.

Connects the hosted Knowra MCP server: search arXiv papers by topic, author or exact UTC dates without an account, and sign in with OAuth from the plugin page to read Scout reports on methods, evidence and limitations.

Session-level lesson-ledger watch for DeepSeek Harness: shows whether a session actually consulted the ledger and recorded the pitfalls it hit, keeps the rule base's health visible (rules needing rework, rules already revised), and can inject the most relevant rules into the step as a system-reminder (high-confidence mode, off by default). The ledger rules themselves live in the external lesson-ledger skill, which the bundled install.ps1 provisions; this plugin only reads them. Zero runtime dependencies; the browser half only needs React.

Adds backup-then-delete and delete-only actions to the DSH session menu: the session directory, its projection cache and a metadata snapshot are copied to a local archive folder and checked file by file, then the originals are deleted to free disk space; subagent sessions it spawned are handled in the same run, and each backup folder keeps a RESTORE.txt with recovery steps.

One prompt row carrying two independently switchable halves. Identity: a fixed self-reference restated at the head, at the tail after every other section, and in the runtime snapshot, with an instruction to treat contradicting claims as stale cache. Discipline: ten rules, each paired with the condition a reviewer checks — external evidence before claiming done, change face equal to review face, correctness from named call points rather than timing, one source of truth, extension through the registered seam, runtime validation only at real boundaries, loud failure instead of silent skips, orthogonal outcomes reported separately, tests through the delivered entry point, and answer shape (conclusion first, length set by the change).

Adds a resizable workspace file tree beside the native preview, with directory expansion, filtering of loaded files, and refresh.

A draggable task-tree panel inside the chat page: capture tasks by voice or text, drag to reorder, edge-snapping collapse, and a plain-text queue file you can edit or version yourself.