Plugins

Connects the Xiaozhi voice assistant to DSH Web over MCP: DSH is the tool provider, weaving 35 DSH Web endpoints into 16 voice-friendly tools for workspaces, sessions, chat, models, settings and files — outbound WebSocket to the Xiaozhi MCP access point by default (no public IP or port forwarding needed), several devices bound at once, plus a DSH settings page with live status.

Bitmap-frame context compression plugin for DeepSeek Harness: renders discarded conversation history into pixel-font PNG frames for vision-capable models.

Markdown visual enhancement and session workbench for DSH Web Composer: live syntax presentation with caret-row source reveal (inactive-line markers can be hidden or faded), deterministic local completions, and list/quote continuation on line break with ordered-list renumbering; long pastes fold into persistent chips with inline context or file-mode delivery (@path across 50+ extensions) plus sidebar LLM rewrites (stop, preview diff, undo/redo); real-time context map with token gauge, composition breakdown, per-turn growth chart with anchor lines, in-place /compact, turn forking, and a Context Doctor workspace (overview, diagnostics, analysis, feedback).

DeepSeek Harness (DSH) 推理等级自动选择插件：由 Jev System One 模型判断每条消息值多少思考量，按模型声明的等级自动推导档位，上下文信封让「继续」这类追问继承话题深度，低置信度向上取，任何失败都静默沿用原等级。 | Jev-driven reasoning effort per message: per-model ladders derived from what each model advertises, a fixed-size context envelope so follow-ups inherit topic depth, ties break upward, silent fallback on every failure path.

Context compression for small-context models: visually configure the compressed context budget, default to an 8k handoff + 16k recent context, and keep full history searchable.

Windows minimal agent preset for DeepSeek Harness: fixed one-line persona, three tools, no runtime context.

Fronts stdio MCP servers with three router tools (list_capabilities, describe_tool, call_tool) so schemas load on demand instead of riding every request; shapes every tool result through a deterministic output ladder that archives the original first (retrieve_spill reads it back); prices tool groups in tokens and can hide built-in groups per agent. Bridges to a standard-library Python router, so Python 3.9+ is required.

Compaction fallback for compaction-basic: when its LLM summary throws, comes back empty or cannot fit the window, the span is compiled deterministically with a seq pointer on every line (vendored dsh-compaction-instant compiler). Adds recall and search tools, /recall and /context commands, an idle trigger at 45% of the window and an on-disk span archive. The engine mounts as one row in the agent preset (docs/dsh-integration.md).

Merges other sessions' context into the current turn: evidence-weighted conflict resolution, retraction of claims whose premises were retracted, trust labelling that resists cross-session prompt injection, and budget-bounded packing. Ships a benchmark.

Distills session work into background observations and reflections to preserve context across compaction.

Feeds every dsh prompt into the local agent-feedback-loop feedback-memory pipeline — captures dissatisfaction, reviews it offline into reflections, and attaches the compiled recurrence-rule context back to the turn, never blocking a prompt. Lessons are compiled into a layered rules block (an index that names every family on every turn, plus on-demand instructions), same-class reflections are merged so one lesson is not re-read as many, and each index fact decays through current → possibly-stale → dormant instead of being trusted forever. Requires the agent-feedback-loop runtime; running agent-feedback-loop install afterwards wires existing dsh profiles automatically. On every prompt it runs ~/.agent/feedback-loop/hooks/core-hook.sh with the prompt on stdin.

On-demand memory for DeepSeek Harness: session logs are distilled into layered markdown (a summary, an index and per-topic files), and a mem_query tool returns only the matched entries rather than the whole index. Nothing is injected per turn, there are no dependencies, and every entry keeps a source pointer back to the conversation it came from.

Multi-language codebase intelligence for MCP-compatible IDEs, the CLI and DeepSeek Harness: chat, search, audit, refactor, and board-ready reports from local code.

Per-workspace MCP scoping for the DeepSeek Harness — decide which MCP servers a session may see from the directory it was opened in.

Workspace drift awareness for DeepSeek Harness agents: detects external/user file changes and injects them into the agent context at step boundaries.

Multi-language codebase intelligence for DeepSeek Harness and MCP-compatible IDEs: chat, search, audit, refactor, and board-ready reports from local code.

人脑式上下文压缩逻辑，大幅减少长任务上下文占用以及token消耗

Policy plugin that gates kubectl by kubeconfig context: hard-deny irreversible verbs outside local clusters, ask for the rest.

Context trimmer: lists the current session's context entries with role, preview and ≈token estimate (cl100k), highlights over-threshold items, and removes selected entries exactly without LLM summarization.

Side-ask drawer opened from a completed answer's action row: follow-up questions run in an independent, auto-archived session seeded with the prior conversation text (read-only, capped at 48,000 characters), leaving the main session untouched. The side session inherits the main session's model, switchable inside the drawer, with a context-usage ring.

Global prompt-injection / context-virus defense for DeepSeek Harness: scans tool arguments, tool results, pre-model messages and the outbound stream; quarantine/block/monitor modes, canary trap, and an evolvable on-disk signature library with learn/import/export.

Turns a Xiaohongshu, Douyin, TikTok, YouTube or X link into text: transcript with timecodes, on-screen text, a description and OCR of every image, caption and metadata. Mounts the hosted MCP server over streamable HTTP.

Cipher memory platform as a DSH bundle: four memory types (attention, experience, preference, cognitive) sedimented per turn by four memory agents, cross-session deterministic recall injected via systemPrompt.context, insight review kept out of user-facing output, per-session UNNI/LOOP modes with memory-driven auto-continuation, memory search/ingest tools, plus a settings page (SOUL editor) and an AgentPool panel.

MCP server that searches the Roblox DevForum and official creator docs so AI coding agents can debug Roblox games against real community and engine knowledge.