Plugins

Persistent Codex advisor threads for DSH sessions, with six native tools, a Codex trail, a sidebar status card, and explicit model/effort provenance. Requires a locally installed Codex CLI with app-server support and a logged-in ChatGPT/Codex account (the CLI is not bundled); tested on Windows 11 with DSH Desktop 0.2.1-rc.5.

Treemap disk usage map with linked cleanup suggestions, a reviewed trash plan, and an append-only audit log on macOS.

Per-Bot memory for Tobs Bots (requires DeepSeek Harness with @botharness/core; it ships off with no Bots listed, and after you list a Bot and set mode to on it works with no other service): a read-only local full-text index of that Bot's own past conversations (Chinese words found inside sentences), memory imported from other AI tools, a status tool, optional auto recall before a step in which a Human speaks, and a service for other plugins. Richer backends can be chosen at install, with a quote first and a daily token cap: a public fact store (memv2-lite), the official Graphiti graph, or both, fed from the Bot's own conversations.

Cross-session long-term memory for DSH: eight memory layers, four notebooks, hybrid SQLite FTS5/vector/graph retrieval, agent-driven automatic capture, write lifecycle hooks, optional write audit, project spaces, forgetting and a visual management page. Full FTS5/vector/graph retrieval needs the bundled Go service, which ships only as a Windows x64 prebuilt; on macOS/Linux, build it from source or the plugin falls back to Node SQLite without FTS5 and retrieval degrades.

A cross-workspace Library for DeepSeek Harness: agents record files, tools, environments, knowledge and work records so any later session can index, search and reuse them.

Adds a knowledge-base sidebar backed by the separately installed local Obsidian Quote service: select registered vaults or folders, retrieve read-only note excerpts and cited context, extract rules, generate project convention briefs, and view shared citation statistics.

Zhinao (second brain) for DeepSeek Harness: Markdown-authoritative layered memory (L1 long-term model gated by user confirmation / L2 knowledge with mandatory sources / L3 recent status under daily quotas), a local recall ledger explaining every injection, lifecycle engine with weekly review and a Web panel. First-run wizard initializes a blank brain, adopts an existing Zhinao directory read-only, or fills in only missing skeleton files; existing files are never touched.

Local searchable archive of DSH chat history: decodes the multi-frame zstd session logs (thousands of concatenated frames), extracts user and assistant text into SQLite, and exposes sync / search / stats tools. Idempotent imports, no external service, nothing leaves the disk.

Elysia's memory plugin for dsh. One shared, human-editable MEMORY.md across every session: two-disk mirror, history snapshots, cross-process write lock; entries tier between the main file and a searchable archive by hit frequency; retrieval is keyword co-occurrence plus BM25. Nine sections (persona, requirements, corrections, lessons, methods, requests, sessions, method library, ingested docs) plus a dedicated self-identity section for a character's own self-cognition, injected above the persona. Ships an opt-in memory pack (packs/elysia: 61,903-char identity + 2,482-char persona, distilled from 1,696 officially-confirmed voice lines and 112,175 dialogue lines — not written by hand, measured) that is off by default, backs up before writing, only touches identity/persona, and can be uninstalled. Also opt-in voice playback through a local GPT-SoVITS engine, a skills library, and a sticker library that sends a fitting sticker at the right moment.

Workspace artifact manager that sorts files into artifact, process, memory and material roles, with extra workspace roots, 12 file types, a year/month/day timeline, preview, rename, trash, zip archive and optional COS sync.

Sync DeepSeek Harness sessions and workspace files to a private GitHub repository for cross-device handoff, with sync badges, search and bulk upload, independent local/remote delete, conflict "pick a side", and multi-device-safe index merging.

Layered context management: at each turn end older tool output becomes a deterministic card, the model can fold older ranges into its own summaries in up to three tiers (compress), and every original is archived locally for decompress and search_context. Works alongside DSH-Mem0: superseded runtime-context snapshots (such as recalled memories) are never archived, carded or folded, and are replaced by a one-line placeholder at turn end when the prompt cache is already broken, when the saving outweighs the cache miss, or above 75% of the window; a strataBudget service (pressure, allowance, countTokens) and a strata/fold-landed event are provided for other context plugins; strata_status warns when DSH's automatic compaction is off or its official summary keeps failing. Its L1 layer, archive rotation, prune guard and preset takeover are forked from dsh-live-compact 1.0.6 (MIT); uninstall dsh-live-compact or billion-context first — they replace the same compaction rows and Strata will not take over while either is installed.

Passive recorder that replays the main agent and every subagent on one timeline, flags repeated calls, loops, recurring errors and truncated output, and writes an AI cause-and-effect review on demand.

Slumber Pavilion physically moves archived DSH sessions out of the load path into ~/DSH-Archive, and ships archive, read, restore and delete tools for them.

Feishu OAuth login reduces the need to distribute the DSH owner token while retaining token login; authorized users receive native owner session cookies with shared owner privileges, without per-user isolation.

Makes DSH's OpenAI Responses calls stateful: after the upstream confirms it stored a response, each turn sends only the newly added input items plus previous_response_id, and a stateless upstream is detected on the first response so every later request falls back to the full history.

Per-provider User-Agent and stream idle timeout settings, with visible request retries and optional bounded continuation messages.

Merges your configured, runtime-registered and locally cached models into one panel, probes and minimal-request tests them, pulls each channel's upstream /v1/models catalog, and chains the keepers into named schemes. A virtual ModelPool provider fails over at request time with per-model cooldowns, exponential backoff, image-request avoidance for text-only models, and a 10-minute memory for rejected reasoning levels. Adds a token ledger by day and model with CSV export, and 21 regression probes. Not a fork of dsh-llm-fallbacks.

Adds a per-session Standard/Fast switch; when the administrator fills in provider, source and target URLs, a loopback proxy injects service_tier priority into matching Responses requests and forwards them to that target.

Runs one task through several configured models at once — in parallel or as a primary/secondary review — and returns each model's answer plus an aggregated digest.

Color and conversation-background presets for DeepSeek Harness, preserving native typography, layout, and interactions.

Eight Material-family colour themes for DeepSeek Harness (Palenight, Ocean, Darker, Lighter, Catppuccin Latte, Vitesse, Snazzy and GitHub Gray), each with light and dark palettes, switched from a row of colour cards in Settings → General → Theme.

Channel test page for any configured provider/model: multi-round prompt-cache hit rate, streaming TTFT and tok/s, and a pelican SVG test.

One click beside the composer turns a half-formed draft into a structured prompt the AI can act on: goal, constraints, output format and acceptance criteria filled in, placeholders only where a real decision is missing. Rewritten in place and never sent — undo with one click or Ctrl+Z. Four editable presets (general / code / writing / translation), optional conversation context to resolve references, and a model that either follows the session or is pinned to a chosen provider and reasoning effort. Requests survive session switches and typing while a rewrite is in flight; a result that no longer matches the draft is kept for retry instead of overwriting it. Pure JS, no build, no bundled dependencies.