Plugins

Context window auto-management for DeepSeek Harness: when the window fills up it rolls over to a fresh one with zero summarization model calls, archiving the old window verbatim for line-by-line recall; handoff anchors carry task orientation, working notes survive compaction, and a stale-notes gate warns instead of vouching for outdated notes. The trigger follows the provider budget equation (window - max_tokens) and the real routed model even after a mid-session switch. One-click composer button plus /smart-compact, three-band automatic rollover, mandatory anchor sanitization; works out of the box.

Intelligent token optimization via Jev System One triage, dynamic tool pruning, and subagent routing.

Model-autonomous context compaction for DeepSeek Harness: a mind_map tool surveys the agent context surface (stable seq ids, roles, token weights, clear-boundary markers, the clear-mind playbook), a clear_mind tool replaces a chosen conversation-history span with a model-written checkpoint through the native compaction transaction, call and result pairs self-collapse into one-line tombstones, and a proactive reminder nudges the model when the context grows long or a turn runs too many steps.

Exports the agent loop to GreptimeDB as OpenTelemetry traces, metrics and logs, with seven Grafana dashboards covering token usage, cost, session outcomes and per-turn latency.

UI harmonizer for DeepSeek Harness: normalizes unfinished or self-contradictory official UI, reconciles style conflicts between installed plugins, and unifies the visual language via official design tokens — with live-customizable chat width, markdown font size, workspace scale, UI font stack and a rounded center-column card from Settings → General; client-only dsh bundle, zero model cost, no host side, removable without residue.

Burned DSH tokens become incense; spend one stick to raise or lower today's case. The case updates daily, Liangwei decides today's Liangzi, and the result is archived at day end.

Connect a ChatGPT account to Codex models in DeepSeek Harness through browser PKCE or device-code OAuth, with automatic token refresh and Web or headless login.

Per-plugin usage budgets and cost caps on the llm/stream waterfall: per-model usage accounting, session and monthly budgets with warn/block transitions, latency windows, and a carbon-footprint estimate.

OpenCode Go usage bar in the composer dock: rolling/weekly/monthly quota rings, DSH-session token & cost stats priced with live official rates, per-model and current-session scope breakdowns with live per-session updates, one-click API key copy.

Cost tracking priced from an online catalogue (models.dev + OpenRouter, 8000+ models, each call priced at its own instant): a per-turn cost line, a Cost tab attributing spend to tool output / model output / system prompt / commands, budget, monthly forecast, ~166 live-rate currencies, and a bill_stats agent tool.

Web UI polish layer for DeepSeek Harness: normalizes unfinished or self-contradictory official UI, reconciles style conflicts between installed plugins, and unifies the visual language via official design tokens; live-customizable chat width, markdown font size, workspace scale and UI font stack.

Extensible animated response pet with blue-whale, orange-cat, and silver-shaded-cat themes, per-session progress, model and tool-call traces, token counts, output speed, elapsed time, and collapsible multi-session status cards.

Five tools for reading a video as raw material rather than as something to watch. video_outline maps a whole file as text and draws nothing at all: 4-60 equal windows, each with its cut count, speech share, intensity bar and tags, then the windows worth opening and the exact start/end for the call that follows -- zero image tokens. video_probe returns metadata and a reading plan without decoding frames. video_storyboard turns a window into one contact sheet plus a text index binding every cell to its timecode, cut count and frame time; cells whose text would be illegible at that scale are marked zoom, and dedupe global folds a picture that comes back after a cut. video_zoom re-renders 1-9 cells from a short window at a larger scale. video_read decodes only keyframes, OCRs each screen, and returns an outline plus \[mm:ss] text -- the path for lectures, slide decks and screen recordings, where the words are the content. It also reads a file with sound and no picture as a transcript (subtitle track, then DSH_ASR_CMD, then a transcript file you pass; long audio is cut at 45 s onto silence), composes six changed frames into one sheet when a file has no readable text at all, and can write the result out as SRT, WebVTT or text. Cuts, dissolves and one-frame flashes are found in a single ffmpeg pass; from ten minutes up a file is analysed on its keyframes only, 2.7x faster on a two-hour lecture; and a cell is never shrunk below a fifth of the source, so a 1440p sheet spends the same tokens on fewer, bigger cells. OCR is the engine Windows ships, tesseract with line boxes, or DSH_OCR_CMD. It takes a B station link (bilibili.com/video/BV..., a bare BV id, an av id or a b23.tv short link) wherever it takes a file: up to 300 seconds is read from the CDN in place with nothing written to disk, longer videos are downloaded once and cached within a 4 GB / 24 video budget, play URLs are cached for ten minutes, and a link that is not a single video says what it is. Needs ffmpeg, uses ffprobe when present, and has no other runtime dependencies. It is a wire-compatible reimplementation of SAOS-3624/dsh-video-storyboard: it registers the same video_probe / video_storyboard / video_zoom tool names, so the two cannot be enabled together.

Read-only session-log projector (drsc) for DSH: scans ~/.dsh/sessions/**, decodes zstd/tiled-frame logs with a vendored leaf decoder and rewrites per-session projections into its own SQLite (node:sqlite, segments / replaces + FTS5), exposed through a single fenced HTTP route — surface ledger, recall/compression pair accounting and Chinese-capable full-text search. Token counts share the exact DeepSeek-V4 tokenizer and formula as CCI verify, falling back to a labelled chars/4 basis when assets are absent. Never writes session files; all output lands in its own data directory. Serves three consumer plugins: context-compression-improved (surface marking + recall reconciliation), session-steward (search-corpus switch), prime-memory (historical anchor backfill). Hosts 0.1.0-rc.2 .. 0.2.1-rc.2 (17-tag exact engines.dsh enum). npm: dsh-reference-session-collector.

Relays the ZCode CLI (the client-bundled coding agent) into DeepSeek Harness as a model provider: a resident app-server session injects the Start Plan account and signs each request with the OAuth plan token, so model requests burn ZCode Start Plan / Coding Plan quota directly — real streaming, image input, per-session resume, no API key.

One-tap connect SenseNova Token Plan language and image models into DSH (off by default), with image recognition and thinking level filled in automatically. On a rate limit it backs off and retries on its own instead of misreading it as an exhausted quota. After signing in to the SenseNova console you can see the usage pill, credit pools and per-model consumption. Xiaohuanxiong can also be signed in by QR code, bringing in its models and their consumption multipliers, and taking over DSH web search.

css-in-js styling kit (styled / Token / Styling) for the forked conversation UI: consumes the official --dsw-* design tokens, so no CSS Modules precompile step is needed.

Web GUI convenience suite with appearance controls, timeline navigation, project folders, prompts, conversation export, Mermaid rendering, quote reply, and DeepSeek balance and daily token usage.

Token usage statistics for DeepSeek Harness: per-model daily and range charts plus a yearly contributions heatmap.

Mobile remote-control suite: bundle plugin with native sidebar entry and admin drawer, self-healing bearer-token gateway over LAN/Tailscale, Android app (sessions/approvals/questions/goals), /fs/* file transfer (Range resume, 2GB upload), latency-based multi-server auto-switch, offline chat cache.

Token, cost, and balance dashboard under the composer: live indicator plus zero-dependency SVG charts for per-turn usage, estimated cost, and DeepSeek account balance.

HUD status panel: Git status, MCP servers, skills, model and token usage in a floating side panel.

Bridges DeepSeek Harness to LINE: runs its own signed webhook server, sets up a Cloudflare Tunnel (quick URL or an API-token tunnel), keeps the LINE webhook endpoint pointed at the current tunnel, and gives each LINE user a separate session with a chosen persona, plus optional scheduled check-ins. Requires a LINE Messaging API channel (Channel ID and Channel Secret); only allow-listed LINE users, or users approved from the pending queue, can talk to the bot.

Connects free and check-in account channels to DeepSeek: Huawei Cloud CodeArts, Tencent CodeBuddy / WorkBuddy, Qoder, TRAE, Raccoon, LobsterAI, MiniMax, ZCode and iFlytek Loomy. Sign in to each channel in settings, claim the daily check-in where the channel offers one, sync the available model pool, and use those models as ordinary providers in the model picker, with a local usage dashboard that counts requests and tokens per account and model. Credentials stay in ~/.dsh/get-it-for-free/ on this machine; no telemetry and no requests to upstream repositories. Its optional LAN relay is off by default and listens on 0.0.0.0 when you enable it.