Plugins

Requirement-driven dsh plugin discovery and environment recipes: plugin_guide matches natural-language needs to a curated plugin directory with reasons and install commands; recipe installs whole setups (通知全家桶, 安全审计套装...) from bundled community bundles with ordered plans.

A dockable IDE workbench for the DSH web UI: a GPU-accelerated terminal (restty renderer over a Rust PTY), a CodeMirror editor, a file explorer, notes, a sandboxed browser, and ripgrep content search.

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.

Anti-bot HTTP fetch tools hardened by SXS's production scraping stack (millions of requests/day): UA-pool rotation, retry with exponential backoff, anti-bot wall detection (captcha / verification challenges) and adaptive rate limiting — tools: `sxs_fetch` / `sxs_fetch_json` / `sxs_rate_status`.

OpenAI-compatible image generation in DSH Desktop chat: generate inline, edit reference images or chain-edit previous results directly (image-to-image) without re-uploading, with storage cleanup.

Command Code Go provider adapter for dsh: scan the Go-plan model catalog dynamically and serve /alpha/generate through the harness LLM seam

Generated-catalog plugin discovery and npm installation marketplace for DeepSeek Harness

Automated unit-test generation: a /testgen command and generate_tests tool that scaffold, run, and fix tests until they pass (LLM + offline template generators; vitest/jest/mocha/node:test).

Collapsible iOS-style stats pill under the composer: session cost, DeepSeek account balance, cache-hit rate, and token usage in a frosted panel.

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).

Generate images and videos, edit images, remove backgrounds, upscale images, upload reference media, and check Fotor credits.

A 护航 panel in the session header that switches the local dsh-guard watchdog on or off and shows whether scanning is enabled, whether the guard is alive and when it last scanned; the Python watchdog that does the watching is deployed separately and found through the DSH_GUARD_ROOT environment variable. The watchdog defaults to a Windows checkout (D:\AI\Projects\dsh-guard); set DSH_GUARD_ROOT to point at another directory.

Explicit, project-scoped memory handoff: when you open a new conversation, one model call that you confirm curates the project's earlier conversations into a bounded brief and injects it into the conversation's first step. No silent background calls; a Memory panel shows the brief and lets you edit notes.

A chibi whale maid desktop pet for DeepSeek Harness who eats white rice while the model generates, lives inside the DSH window or anywhere on a macOS or Windows desktop, and dives for treasure between turns. On Windows the desktop overlay is compiled at install time and skipped when swiftc or the .NET SDK is unavailable, so the pet stays inside the DSH window.

Turns a phone into a personal assistant for the DeepSeek Harness on your computer — pair once by QR code, then chat, send photos and files, watch replies stream, and collaborate from several phones at once over LAN, Tailscale or a public tunnel. Ships an Android app, an iOS build pipeline and a zero-install web/PWA client for HarmonyOS and iOS browsers.

Local non-generative decision layer: phocinae_ask and phocinae_gate return typed verdicts with calibrated confidence from a small decision model, with fail-closed escalation. Needs a local phocinae-server (pip install phocinae-server) with the Phocinae-Largha-150M-v1 model on loopback port 8155.

Adds Regenerate and Edit and resend to each message you already sent: Regenerate re-answers the same text in one click without going through the composer; Edit and resend retracts that message and everything after it in place — out of the transcript and out of the model context, in the same session with no new branch — then puts its text and images back into the composer, so the edited send rewrites the conversation from that point. Messages carrying files rather than images stay disabled.

Forensic analysis of past model-routing decisions. Replays each session, classifies every turn's task complexity from its request shape (deterministic, no LLM), and reports two kinds of waste: an expensive model on cheap work (the difference is measurable), and a cheap model on expensive work that failed on its own merits — where every retry re-sends the whole context, so the downgrade cost more than the upgrade. Distinguishes provider-capacity retries (RATE_LIMIT/TIMEOUT/TRANSPORT) from model failures so it does not cry wolf. Derives routing rules from the evidence, priced against a built-in per-million table.

Visual reference moodboard for DeepSeek Harness: drag-and-drop images with tags and notes, and pinned style guides the agent reads before it generates or restyles UI.

Composer UI for DeepSeek Harness that does two things. The reasoning-effort row in the model menu becomes a continuous slider drawn on the row itself, with the rungs read live from the model adapter so any ladder length works; each rung paints its own ember plume into a canvas behind the knob. Separately, the permission trigger renders in the theme error colour whenever full access is the active preset, including its row in the permission menu.

Four-tier prompt enhancer for the composer — the system tier always emits a LangGPT-structured system prompt (# Role / Profile / Skills / Rules / Workflows / Initialization), upgrading task drafts into the system prompt of the agent that runs them, while the other tiers rewrite as a light tidy-up, a single paragraph, or a five-section brief; the iterate workbench refines an editable base under an optional requirement, with fill-back and undo. Never changes intent, never invents requirements.

Composer meter for the DeepSeek Harness Web UI: a pill under the input shows draft character count, estimated tokens and this session's estimated cost from DeepSeek's published prices (peak/off-peak, Flash/Pro or custom rates), and opens a panel with the token breakdown, a six-snapshot per-session draft vault and a draft cleaner; every write goes through the host setDraft API instead of touching the editor DOM.

Token-usage profile for DeepSeek Harness — a Codex-style usage panel inside Settings: lifetime tokens, a year of activity, cache hit rate and top models. · DSH 用量统计面板：在设置里查看累计 token、一年活动热力图、缓存命中率与最常用模型。

Sidebar tab that lists a session's file changes by turn, expands line-level diffs and undoes them, with a matching review card at the end of each chat turn. Needs dsh-better-sidebar installed separately. DSH 0.2.x port of Lzh3070/dsh-file-review-tab; do not install both.