Плагины

AI companion with a learned body clock, real offline sleep, its own memory, and self-reported affect delivered through the same judge math as the pi version.

Optional JEV decision mode with deterministic project reads, bounded semantic selection, native authorization checks and desktop credential settings.

OpenCode Go in the DeepSeek Harness model list, providing subscription quota display, model discovery, session affinity, and more.

Token usage panel for DeepSeek Harness: a bar above the composer opens billing token totals, trend charts, per-model and per-project breakdowns and a custom date range, read from local session logs of DSH, Codex, Claude Code, Trae and WorkBuddy. With a self-hosted portal URL and an app key it also reports usage events to that portal; with neither configured it collects nothing. It exposes token_usage tools plus a ctx.tokenReport service, and never collects conversation content.

Refines the built-in cache-hit percentage to three decimals in place across the composer stats row and the token-usage dialog.

Live usage gauge injected into the system prompt every turn: session tokens and estimated spend, a next-turn estimate, and two predictions the model can act on — which step fills the context window and which step burns the budget. /qm panels, zero-config by default, optional price table, and it publishes a summary.json budget contract sibling plugins read.

Prompt optimizer for the composer: a button in the tool row rewrites your draft through a model channel you already configured, streaming into a panel that floats above the composer without taking up layout. Your draft changes only when you click Accept — and Accept is refused while the answer is still streaming, so a half-finished reply can never be written back. It asks first when the draft carries @ references or attachments, which a whole-draft replace would flatten into plain text. No global keyboard shortcuts, and no API key ever reaches the browser: the host half makes the call.

DeepSeek Harness plugin: show a provider + model icon in the sidebar for every session, updated live from the composer model selector with a last-request fallback for history sessions; aggregator routes render the provider logo with a 10px model logo in the corner.

One-button prompt optimizer: rewrites the composer draft via a configurable strategy, with an editable template, few-shot examples, pinned models and reasoning effort.

Compact 56px app navigation rail with a separate conversation list and sidebar collapse control for the DSH web UI; also hides embedding, rerank and image/video generation models from the chat model menu and /model.

Adds a Chinese-mode toggle beside the composer that opens a panel for switching reply, reasoning, and tool-narration language, persisted in the profile.

Deterministic goal-completion gate for DSH: a contract-first AC grammar with named failable check commands, a gate that re-runs every check itself (zero model trust), verdicts bound to a tree digest so file changes void passing verdicts, and a hard deny on false completion at update_goal via tools/pre-execute with a false-complete counter that blocks after two catches; scope-aware task-level claim handling (strict or goal-only) keeps multi-task Agent Teams loops from deadlocking.

Project-level memory built around a ledger whose structure the AI designs per project; the plugin only handles detection, injection, write-back, archiving and git. Memories are split by meaning: the active ledger holds current state, the turn journal is append-only, and decisions, the goal tree and lessons are one-card-per-file (with a multi-dimension index over the card folders); the fundamental principles file can be changed only by the user. There is a single compaction mode: before compacting, the AI sorts the work into three buckets - what must stay in context and is re-injected verbatim, what is written to the project's _stash/ for later lookup, and what is dropped. When a conversation is opened in a container workspace holding several projects, it can be locked to one sub-project (the lock lives in the plugin's own home and survives restarts; it leaves no marker in your project). Also an activity log with five levels, where L4 stores the user's own words verbatim (never injected, queryable on demand). Before each compaction a verbatim capsule (user quotes, raw errors, changed files) is written to disk, and both the capsule and its source material survive a restart; a handoff document lets a new conversation resume the previous one. 21 tools, 12 subcommands, 39 settings (one of them switches whether the settings panel or each project's manifest wins). Runs locally, no runtime dependencies.

Cross-compaction memory for a long session: on compaction it stores what is being dropped, and at question time it injects a few relevant passages (or nothing) so the model can recall what you decided before.

Cross-session message forwarding between DSH sessions: extract one model reply, a user message, or a multi-selected batch of messages and images from one session and deliver it into another, with silent send, draft fill, or send-and-jump delivery, plus prompt templates.

Replaces the composer model selector with a "model · reasoning level" trigger: a slider across every level a model reports, three switchable modes (the DSH native selector, a gradient particle rail, an energy-charge track), and a per-provider one-click fill for models that declare no reasoning levels.

Reports which PowerShell the DSH pwsh executor rows actually run — enabled rows only, with version-numbered Store paths refused as pins — and installs the self-updating Microsoft Store package as a background job after reading the connection cost, because Windows defers Store downloads on a connection marked as metered. Two tools and a prompt section on the host side, plus a configuration card, a diagnostics section and a re-detect action in the Web UI; Windows only, no runtime dependencies.

Registers the TypeSafe/Jev (System One) skill into DeepSeek Harness's global skill layer, so every session can classify, route, score, extract and gate on confidence using typed judgments instead of prompt-and-parse. Ships no API key; the user supplies OPENROUTER_API_KEY.

Composer and sidebar helpers in one plugin, each switchable in Settings: a git Changes panel (uncommitted / unstaged / staged / branch diffs), ↑/↓ input history, Shift+Tab /plan toggle, one-click ▶ Continue after a stopped or failed turn, drag-and-drop queue reordering, Copy deeplink, Auto Review "Allow for this chat", project status markers with active projects first, background notifications, 11 extra UI languages and automatic updates from GitHub releases through the plugin manager.

Translates DSH's private remote protocol into a versioned REST API (chat, sessions, files, devices) for trusted clients such as the Android app dsh-companion, plus a Phone-settings panel with seven inline cards inside the DSH WebUI settings.

Three-tier deny/ask/auto gate for the computer-use (cua-driver) tool family: each tool category gets one segmented control in Settings, and an allow list or deny list overrides the tiers per tool name.

Notifies you about 11 configurable DSH agent events (task completion, questions, denied approvals, background subagents/goals/jobs, team tasks, context compaction, triggered schedules, plugin startup and permission changes) as browser Web Notifications through a bundled service worker, falling back to Windows toast or Linux D-Bus, with three silence levels and click-to-open.

Synchronizes DeepSeek Harness third-party plugin lists through S3-compatible storage with diff preview, local snapshots, rollback, and optional LLM provider configuration sync.

Reads the session event stream, with no extra model calls, and turns corrections, failures and fixes into pending lesson candidates. Every candidate is judged against written criteria and carries the reason it passed or was refused; only confirmation writes a skill, and only through the single validated writer `learn_skill_manage`. An idle curator marks unused skills stale and archives them by moving rather than deleting, and spares any skill that is still being loaded. Five tools: `learn`, `learn_review`, `learn_curator`, `learn_skill_manage`, `learn_skills`. Registers a prompt section stating the criteria, and a tool guard that refuses direct edits to skill files so the writer cannot be bypassed.