Plugins

Adds 17 barotrauma_* tools to DeepSeek Harness for making Barotrauma mods: environment probing, mod skeleton, item and weapon XML, afflictions, creatures, random events, Lua and C# script scaffolding, localization, validation, icon generation, zip packaging and Steam Workshop publish notes. barotrauma_gen_icon sends your prompt to image.pollinations.ai and writes the returned image into the mod folder; on Windows the Steam install path is read with reg query.

On-demand MCP tool loading. Only the mcp__* tools listed in the coreTools whitelist (max 5) stay in a session's tool set; every other MCP tool definition is actually removed from the request through the session-scoped tools.restrict() API and comes back only after the model calls lazy_load_tool. A tools/pre-execute gate turns calls to unloaded tools into a load-it-first error, and activation resets when the session ends.

Local-first memory nebula for DeepSeek Harness with graph views, evidence chains, hygiene checks, and offline snapshots. Reads the memory stores of every workspace and of other memory plugins (read-only, for example meow-memory's SQLite) and writes only under $DSH_HOME/memory-nebula. Registers 9 agent tools and a system-prompt section (usage rules every turn, plus not-yet-read neighbouring memories after a few reads); its /export and /share endpoints return the full memory text without authentication on the local server, and an optional standalone engine listens on 127.0.0.1:7890.

Edits a message you already sent and re-runs from that point by forking the session, and retries a completed turn, in the DeepSeek Harness Web UI.

Pause or resume the running task from a 28 px button beside the DSH composer: safe pause lets the current step finish, holds the next agent/pre-step boundary, and resumes the same turn without a continuation message; force pause cancels immediately, and the default granularity is configurable in Settings.

Edit and resend or regenerate a message you already sent, including while its answer is still streaming: the plugin stops the running turn through the official Session.cancel, waits for the session to go idle (8s cap), then rewinds that message in place the same way dsh-redo does. Command is /relive so it can coexist with dsh-redo.

Session-archive layer for DSH. It injects one per-workspace Markdown file into the system prompt on every turn, searches past session transcripts with AND keywords, an optional time window and a workspace filter, and recovers the images and documents those past sessions carried by returning their original local paths or copying them to a chosen directory. It also reports index health: version, entry count, cache usage, frequent sessions and sessions stored in an archive format it does not recognise. Keyword search only, no embedding service.

Manual session handoff: when you decide a session has run out of usable context, one click turns its event log into a compact handoff document (goal, done, next, project facts), creates a new session seeded with it and wakes it to continue; the new session can inherit the session-bus name as <name>-handoff. No threshold, no timer, no automatic trigger.

Keeps a root turn open until progressing child agents and the current-turn background jobs it started settle, then asks for one self-contained final answer in that same turn. Idle children that still have parked input do not delay closing. It observes job settlement without consuming job output, never cancels children, and does not decide task success. Verified on DSH 0.2.2-rc.2 only; on any other host version it stays inactive, logs one warning, and turns close natively. Each batch of results that arrives after a reply costs one extra model call; ordinary questions add none.

Folds older oversized tool results into spill files and replaces them with short, deterministic stubs at step boundaries, with no LLM call: the stub keeps the first 200 and last 80 characters plus the sequence number, tool name, line/char/size counts and the original file path, and each fold is a real surface replacement accompanied by compaction/prune shadow pricing so token counts stay honest. Triggers on context pressure at 25% of the model window (falling back to 32768 tokens when the window cannot be resolved) or on five accumulated candidates; the last three turns, error results and rich content are never folded. The original is recovered from the session log rather than the spill file, so /folds, /unfold <seq> and /unfold all keep working after spill cleanup; session_fold_recall lets the model fetch a folded result by sequence number or by keyword, and session_fold_diagnose reports the effective settings. /spill-prune removes orphaned spill directories, preview only unless --apply. A Settings page exposes 12 of the 13 options as draft-then-save fields; they are volatile, so saved changes apply without restarting the plugin.

Puts a domain bar under the session header — a domain groups skills and tools and toggles as one block, so one click loads or unloads every member. Hovering a chip lists its members (skills, then tools) in a height-capped panel that scrolls. The right end of the bar is a compaction-threshold counter (− 5 +, 100k tokens per step) that rewrites the live compaction engine policies, so the compaction trigger line changes on the next turn without restarting dsh; a degraded fallback turns the number amber and is reported in the snapshot. Settings maintains domains and members with search, skill/tool tabs, all/selected/unselected filters and a copyable AI grouping prompt. Tool removal uses tools.restrict plus a guard; hiding a skill re-registers it with modelInvocable false, so it leaves the model catalog while /name still works. Sessions of another preset are left untouched.

Read-only session dashboard for DeepSeek Harness over the local ~/.dsh data: a sidebar overview of every session (tokens, estimated cost, turns, tool calls, model mix, daily activity, searchable and sortable session table) and an in-conversation Dashboard view that shows the current conversation turn by turn with per-turn tokens, cost, timing and tools. Zero dependencies, no network.

Switches which assistant persona answers, based on how the message addresses it (alias matching, with names inside quotes and copula phrases such as "is/was" excluded); each persona is a folder holding its own prompt, aliases, colour, avatar and optional lore, several personas can answer one turn, and a settings panel plus a per-turn routing log make each switch checkable. Personas live under ~/.dsh/personas (deleting one in the panel removes its folder) and the routing log in ~/.dsh/persona-router keeps the first 60 characters of each message; its HTTP routes (/persona-router/…) have no DSH access check.

Qwen mode for DSH: manages local GenieX and Alibaba Cloud QwenCloud providers, normalizes thinking and sampling, detects model capabilities, and shows per-turn latency and decode speed. Its bundle patch replaces the permission plugin's preset table with reply-only, read-only, workspace-write and danger-full-access (the last two as in stock), so any custom presets are overridden.

Dark mecha wallpaper theme with graphite glass panels, drifting particles and a pointer-following spring mesh, with 16 adjustable settings and a one-switch return to the native look. The bundled Gundam wallpaper is not covered by the MIT license.

Arknights-style procedural opening film and silent looping chat backdrop for the DeepSeek Harness Web UI: Canvas-drawn visuals, synthesized WebAudio sound, click-anywhere skip, cross-dissolve into the interface, live-adjustable opacity; zero dependencies and no media files. By default the film (about 12 seconds, with sound) plays on every page load; a setting turns it off.

Adds a recap card under the composer for the current session: turns, steps, tool calls grouped by name, the uncached / cache-read / cache-write / output token split, the cache-hit rate, and a cost range computed from the official Flash off-peak and peak list prices. Exports the same card as Markdown or a standalone HTML page.

OpenCode Go allowance bar under the DSH input box: the rolling, weekly and monthly percentages still left, plus a popover pace budget that turns each window into a per-day and per-week spend target and marks whether the current burn is ahead of that target or behind.

Gives the agent its own context usage through a context_usage tool plus one reading line per turn, with an editable template.

Per-task cost dashboard and budget guard that prices each model call at peak or off-peak rates, turns the remaining budget into a per-turn work envelope in the system prompt, and stops the task at its cap until the user replies. The default cap is ¥0.5 per task.

Turn the seam between the sidebar panel list and the workspace list into a divider you can drag, snapping one row at a time, with the panel list scrolling when squeezed.

Adds a message queue to the sidebar foot: park a thought in a collapsible list, then send one entry to the current session when you choose instead of interrupting the running turn. Entries can be edited or deleted, the list holds two rows and scrolls beyond that, and an unsent composer draft is moved into the queue before a send rather than overwritten.

A macOS screenshot button in the composer tool row: the scissors split button runs the native screencapture marquee — optionally hiding the Harness window first, chosen from its caret menu — then copies the PNG to the clipboard and attaches it to the composer draft; Screen Recording is preflighted, so an un-granted capture points at the settings instead of returning a wallpaper-only image.

Reads EPUB, TXT, Markdown and HTML from the workspaces registered in dsh in a right-sidebar tab, restores the last reading position, and collapses the whole sidebar with Esc to return to the conversation. No runtime dependencies and no build step.