Plugins

Composer-dock balance/usage readout for DeepSeek, Kimi Coding, OpenCode Go, Zhipu, MiniMax, OpenRouter and OpenAI Codex — provider-based, 2s polling, 5min cache.

GEO brand audit skill bundle: AI-search visibility, recommendations, citations, competitor presence and content-gap diagnosis across 17+ AI engines (ChatGPT/Perplexity/Claude/豆包/DeepSeek/Kimi/文心一言).

Compares coding agents in isolated Git worktrees with deterministic validation, scoring, and explicit winner application.

Analyzes conversation images through mode-specific prompts (caption, UI, document, grounding, topology, etc.) and injects structured evidence with coordinate primitives (boxes, points, refs) as text, with session-level caching for reuse across replay and compaction.

Auto-switches the agent to ultra-compressed caveman-style output during DeepSeek peak-pricing hours (Beijing 09:00–12:00 / 14:00–18:00, price x2) and shows a status badge plus transition toasts.

LoongPort DeepSeek Harness (dsh) Cordis bundle and setup CLI

Lightweight OMDSH Runtime and Profile Pack toolkit for auditable DSH composition, signing, candidate installs, and rollback

Delete one complete closed turn from a DeepSeek Harness session without deleting the session

Settings section for archived sessions: browse by workspace, search, batch unarchive with the sidebar updating live, and optional auto-archive rules (idle days, keep-last-N per workspace) with a dry-run preview — plus a left-edge turn rail for jumping between turns of long conversations and ↑/↓ input history in the composer.

Linux kernel dev domain plugin for DeepSeek Harness (dsh): verification pyramid L1-L4 as model-facing tools + graph-layer governance (anti-Goodhart / blind-spot / patch-conflict) + AGENTS.md contract as a system-prompt section. Plain JS, no build step.

DeepSeek Harness plugin: import models with capabilities from an OpenAI-compatible /v1/models listing (gateway capability extensions) and edit per-model capabilities (thinking, vision) on its settings page

DeepSeek Harness plugin for SCNet HPC: SSH setup, CPU/DCU Slurm jobs, cluster probing, compute-node diagnostics, and Hygon DCU/DTK guidance.

Mobile adaptation for the DSH Web UI: compact typography, full-width composer with caret fix, drawer sidebar with overlay-close, two-page settings flow, responsive tables and stats, safe touch targets.

Conversation header badge showing the DeepSeek Harness version, checking npm for the latest release and prompting upgrades.

Plugin compiler with a blueprint registry (scaffold → validate → deploy) exposed as a runtime service, three model tools, and a blueprint-gallery settings page.

Shut down Windows automatically after DeepSeek Harness tasks complete — a floating arm button plus a cancellable countdown banner (shutdown /a), zero runtime dependencies.

Minimal prompt x Web PTC capabilities: one-line persona, run_code SDK, subagents, Ralph, web search/fetch, and Windows Git Bash; the general-purpose workflow tool stays hidden.

Chat image-attachment bridge with a `view_image` tool for any OpenAI-compatible VLM (local Ollama or cloud): pasted/dropped images become `view_image` path markers before reaching text-only DeepSeek models.

Vision provider route that transcribes attached images to text through a configurable model (15+ OpenAI-compatible and Anthropic vendors) while DeepSeek keeps answering.

Vision bridge for text-only DeepSeek routes that analyzes attached and local images through configurable OpenAI Responses, Chat Completions, or Anthropic Messages endpoints while leaving image-capable routes native.

Vision toolkit for text-only DeepSeek: model-invokable `vision` tool, wrapper adapters for deepseek/opencode-go (v4 flash/pro), Antigravity IDE quota (default, flash/pro) / any OpenAI-compatible VLM / Gemini / local Ollama channels, evidence memory with compaction rehydration, content-hash cache, and a bilingual client panel.

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.

Renders the built-in cache-hit percentage with three decimals and widens the composer stats line to prevent truncation.

Bi-temporal memory backed by a SodaMem daemon: recall is contributed to the prompt while each turn is assembled and every closed turn is ingested, with no tool call and no model call on the read path; each recalled fact cites the turn it came from and carries validity dates, so superseded facts stop being retrieved.