Plugins

Cross-session memory engine for DeepSeek Harness: SQLite store + human-editable Markdown mirror, autoDream consolidation, offline semantic search (local vector/rerank/clustering), entity-attribute-timeline, Sleep Mode, custom-model autoSummarize, and 7 memory tools.

A Multitask agent preset where the main conversation only plans, briefs subagents and summarizes results, while file reads, commands, search and builds run inside subagent contexts; four worker kinds, an approval gate for dangerous operations, and zero changes to DSH source.

Deep reading & summarization for DSH: books/papers/videos/web into structured Obsidian notes via plugin parsers, MapReduce parallel deep-read subagents, JSON Schema outputs and an idempotent cache.

Project memory for dsh with native task-system integration: session todos and file reads auto-condense into cross-session task records (task↔file links), so workflows are switchable and resumable without re-contextualizing. Read-time memory (no upfront scan): extracts code symbols, summarizes docs, and cross-links docs↔symbols; optional TS semantic enhancement (auto on when TS installed). Experience notes auto-dedup. Every recall includes a traceable citation. Pure JS, single dep, works out of the box, silent operation, zero overhead.

Smart session titles for DSH: after each conversation turn an independent auxiliary LLM call summarizes the full user+assistant transcript into a title that follows the session's real topic instead of echoing the first message; the first message is titled instantly, failed built-in titles auto-retry on later turns, manual renames are never overwritten, and subagent/fork sessions are skipped.

Replaces the built-in DSH compaction summarizer with a five-section handoff summary and keeps every shadowed message readable in a segment directory.

Multi-agent roundtable discussion: each member runs as a subagent on its own model and speaks in turn, a host summarizes each round, and termination writes meeting-minutes Markdown.

Sign in to DeepSeek Harness with a Claude Pro/Max membership (no API key), downscale Anthropic request images to a 2000px long edge, slim /compact summarizer payloads, and show 5-hour/weekly plan usage plus DSH message counts.

Doubao-style voice chat for DSH: press-and-hold mic in the composer converts speech to text and auto-sends, and AI replies are read aloud. Optional LLM condensing (long replies summarized into short spoken lines, following the current conversation model), TTS-friendly text cleaning, selectable Edge TTS voices, adjustable silence auto-stop, and settings embedded in DSH settings dialog.

Select text in any Windows app to explain, summarize, translate, or read it aloud with DeepSeek Harness — answered in a streaming always-on-top overlay with a horizontal capsule menu.

Preview conversation context and selectively keep, summarize, or omit exchange groups before applying persistent compaction through Harness session services.

Reformats dsh session titles as "date | type | topic" (type and topic are summarized by a model from the full conversation), and hides unused sessions from the sidebar without deleting them.

Context manager for DeepSeek Harness: auto-summarizes every turn into priority-sorted records, injects them plus pinned cross-session notes and custom text into each model request, and folds any-size real conversation ranges into summaries to save tokens. Persistent and restart-safe. NOTE: manual multi-step install — copy the three package folders into profiles/node_modules and add the provided cordis.patch.yml rows, then restart DSH; not a standard one-command install.

A green interrupt button beside the send box: silently stops a running agent, which then summarizes its progress and asks for your new requirements, with a customizable interrupt prompt.

Windows only. Bilibili video summarizer: downloads the audio track, transcribes it locally with faster-whisper (Python + faster-whisper, about 725 MB of models on first run), then writes the analysis markdown and renders a one-page infographic PNG.

Model-callable compact_agents tool that force-compacts context in every live session (main session, ordinary sub-agents and AgentTeams members alike), queues busy targets and compacts them when the turn ends, reports the shadowed node count and estimated tokens per target, and ships a browser form for the trigger ratio, retained ratio, controlled-phase output budget, auto-continue budget and an end-of-turn pre-compaction ratio that compacts near the trigger line once a turn ends, so a reply no longer opens with a wait for the summarizer.

Single-file persistent companion memory: interests, preferences, tasks, and past works summarized into the system prompt, with read/update tools.

Hand long sessions over to a clean one: /handoff summarizes the conversation with an LLM, creates a new session+agent in the same workspace, injects the handoff document as the first message, and auto-opens it; the origin session stays unchanged.

A 轨迹解读 (trajectory interpretation) tab that summarizes each user round — what was wanted, how the assistant thought and executed, files/commands/errors — via a rules engine plus optional LLM narrative; user messages stay verbatim.

Hand off focused context without waiting for or interrupting the active Agent: freeze a bounded Session snapshot, summarize it asynchronously with an isolated no-tool LLM, and deliver a provenance-bearing capsule to another Session or an authorized Room—quiet by default for Sessions, with explicit wakeup available.

Saves architecture and validated plans, runs native AgentTeams workers with approval, parallel coding and independent review, and summarizes progress and verification results.

Notes-backed context compression for DeepSeek Harness: replaces LLM-summarized compaction with a model-written persistent notebook plus recoverable raw history, so the rollover boundary itself costs zero model calls.

Chunked compaction for oversized sessions: takes over only when the live context exceeds 85% of the model window (or the request exceeds 4 MB), summarizes the history in 1.5 MB chunks and merges them; smaller sessions keep the built-in compaction.