Plugins

Dot-matrix backdrop for the DSH Web UI: the dots form the DeepSeek whale or a transparent PNG you import (alpha-masked), each dot carries its own low-frequency force and follows the cursor in manual mode, while an auto screensaver mode sweeps 360-degree waves and full-screen breathing across the artwork; sliders and toggles live under Settings > Plugins.

Unified real-time quota dashboard: one collapsible glass panel for OpenCodeGo windows, DeepSeek balance, OpenRouter credits, SiliconFlow and Moonshot balances — auto-detects DSH credentials, host-side fetching (keys never reach the browser), custom providers via ~/.dsh/dsh-quota-hub.json.

Context-window attribution for DSH Web: a Mini Utility Dock panel that ranks the user messages, assistant turns, reasoning blocks and tool results of a session by the tokens each one occupies, so it is clear which entries hold the window. It also shows the share by entry type, the heaviest entries across the live sessions of the host, and, where the host provides them, provider usage and the estimated system/tools/messages composition. The panel is read-only: it does not estimate cost, modify a session, or trigger compaction.

Teaches Herdr about DSH panes: reports agent working/idle/blocked state — including the blocked state DSH itself lacks — enabling panel jump and --wait. (Herdr integration / agent state display / panel jump)

All-in-one combo patch for STARDUSTLC's 18 DSH plugins: one command injects shared default configs across office, media, DevOps and notification presets; each component installs directly via a one-line add command (flat design, no git subdependencies).

Intercepts turn-stopping to force one critique-and-improve round after each completed turn before the agent stops.

Plan-and-execute orchestration over ReactLoopAgent — turns a goal into reviewable steps with a web UI plan card.

Captures every plan the agent presents through exit_plan_mode as markdown under <workspace>/.dsh/plans, opens a review window, and keeps a Plans history tab recording the outcome of each plan.

Input redundancy pruning for DeepSeek Harness — drop repeated/redundant tokens before they reach the model (layer ②).

Vision bridge for text-only DeepSeek models: transcribes attached images with deepseek-v4-flash-vision-exp before they reach deepseek-v4-pro, with no third-party dependencies.

Route every host-process undici fetch through a local HTTP(S) proxy while keeping localhost and China-reachable APIs direct.

Codex-shaped coding tools and agent presets for GPT models on DeepSeek Harness: exec_command, write_stdin, apply_patch and view_image, plus OpenAI Chat Completions and Responses API routes, bridged to DSH-native plan, ask, approval and compaction. Long-running commands become DSH background jobs with bounded head/tail output readable through job_output, and each subagent type can be pinned to its own preset, provider, model and reasoning effort.

Model Health panel in Settings: lists every model configured in settings.yaml and batch-tests availability and latency for each.

Input prediction for the DSH composer: your next prompt suggested as ghost text after each turn, accepted with Tab. (autocomplete / input prediction / prompt suggestion)

Layered syntax highlighting for bash/pwsh command and read code output in the DSH web GUI (VS Code Dark+ palette: keywords/strings/numbers/functions/comments each get a color); colors only what it recognizes, keeps tables/logs plain; pure frontend, zero token overhead.

Always-on-top capybara task reporter for DSH Web — a transparent floating pet that narrates each trajectory step in a speech bubble, names the project it reports for, and alerts you when a task completes while you work elsewhere.

Adds deliverable-file buttons to the end of each agent turn (opens files with the OS default app) and a floating dot rail for jumping between conversation turns.

Keeps a local model's context at a size the GPU handles well: measures prefill speed on every request and, before each step, warns or compacts early on a hard token ceiling, an observed time-to-first-token budget, or a predicted cold-prefill budget; /context-budget shows the numbers and the cost of compacting now.

DeepSeek Harness Web plugin for visualizing, arranging, branching, merging, and summarizing AI agent sessions on an interactive graph canvas

AI-generated curiosity & trivia plugin for the dsh web GUI (Curio in the English UI): 「开始探奇」 generates a fresh batch of unexpected knowledge (excluding everything already seen), each item unfolds into two deeper layers plus similar knowledge points, and 「探奇清单」 keeps the full history (batch, items, layers, similar points) in browser storage. Reuses the host LLM channel with zero config; when no model is available, a DeepSeek API key can be pasted into the panel. Output language follows the DSH UI language (zh/en).

Percent-progress scoreboard docked below the todo pill: tracking_write rows with evidence-bound percents, host-captured git checkpoints with before/after deltas, and pursue/align/dismiss actions that reach the agent as instructions.

A full-screen, Matrix-styled TUI frontend for DeepSeek Harness — built with React + Ink as a Cordis bundle plugin

Tiered compaction hub: replaces the single 80% compaction cliff with a configurable ladder (default 30/50/70/90) whose tiers escalate from gentle to maximum consolidation laws, folds a deterministic keyword index into every checkpoint so facts accumulate across cycles, and summarizes through a model chain — primary plus up to 3 fallbacks, each with its own reasoning effort. Adds a Compaction module sub-tab and a Settings card; toggling off always restores stock compaction.

Control panel for local LLM inference servers in the DSH web view: service registry with health checks and stop controls, named parameter profiles for llama.cpp, SGLang and vLLM with pre-save KV/VRAM validation, GPU detection, and tok/s benchmarks.