Plugins

Turns the composer's reasoning-effort menu into a draggable slider and takes over the stock model-and-reasoning-level chip. Dragging follows the pointer continuously; releasing snaps to the nearest level and submits one session update per drag. The track draws its own fill, tick marks and white knob, only the highest levels (xhigh, max, ultra) get a flowing brand-colour gradient, and both ends carry the lowest and highest level names. The same card expands into the official model list. Arrow keys step one level, PageUp and PageDown two, Home and End jump to the ends, and each move submits. A generation counter on the directory object discards late responses, and reduced-motion turns the animations off. All data comes from the stock modelDirectories directory; no new capabilities are added.

Reload a plugin or skill you just installed without restarting DSH. reload_plugin clears that plugin's own module cache, unloads the old instance and re-imports it, restoring the previous module if the new code throws; reload_skill forces skill directories to be re-discovered. Ships both as tools and as the /reload-plugin and /reload-skill commands.

Project-less chats for dsh: a session created without a workspace gets its own directory under ~/Documents/DSH/<date>/chat-NN, reused whenever the chat is reopened and left in Ungrouped rather than made a sidebar project; the new-session hero gains a hover clear badge and a "work without a project" row, and a client-only sentinel workspace keeps that hero and its composer usable.

Keeps a CURRENT_PROGRESS.md file in each session's working directory: asks at session start whether to create it, reads it instead of asking when it already exists, and appends one entry after every turn.

Replaces the built-in bash executor so every command gets an environment resolved from its own working directory: configurable PATH prepends, Python venv activation and mise shims, with a dedicated section in the Web UI settings.

Pack a directory into a deterministic tar, read an archive's contents without extracting it, unpack while refusing every member that could reach outside the destination, and verify what came out against what went in by SHA-256. No external dependency — the tar format is written and read in this plugin. Extraction is deliberately more permissive on reading than on writing, because a reader's job is to report what an archive says, while a writer must never emit what it cannot read back. A tar has no central index, so the walk tracks whether it reached the end-of-archive marker: without that, a half-file checksums fine and reads as complete.

Permanently delete a session from the session row menu — archives the id, removes the log directory and drops the row at once, with a Delete shortcut; built only on shipped extension points, and a session that was merely opened stays deletable.

Load each workspace's direnv environment into every command the DeepSeek Harness runs — automatically, per working directory. 让 DSH 每个工作区的子进程自动带上该工作区的 direnv 环境。

DSH host plugin: confines agent scratch scripts to one directory per working directory, and deletes them when each turn ends.

Cross-session lesson book (the agent wrong-answer notebook) for DeepSeek Harness: injects a memory index capped at 2 KB verbatim into every turn, adds mem_recall and mem_save model tools whose write gate demands four sections, a locatable reference and explicit human approval before anything reaches disk, registers the /memory command, and keeps every lesson as a zero-dependency Markdown entry in the .memory/ directory.

The Space Lens workbench as a DeepSeek Harness panel: the directory the current session is working in, scanned and explored by Space Lens itself.

Chain-of-thought panel for the DSH Web right sidebar: folds the live reasoning stream of each session into turn → step groups carrying the reasoning text and the tool calls made in that step. Only the newest turn is expanded by default and earlier turns collapse to one line; steps are grouped into activity phases by a local rule, the step waiting on a tool is marked, tool failures show their reason, full reasoning text is fetched per step on demand, an all-turn directory searches turn titles, user text and turn numbers locally, and a whole turn can be summarised into a Chinese title with one model call.

Session-aware mesh relay for DSH agents: an Ed25519-signed receive route and delivery policy, plus outbound sending, a peer directory and a mesh-peer-registry client.

Enhances the native right-sidebar Files tab rather than replacing it: each row gets an @ reference button that inserts the canonical path mention into the composer, plus a size column, a sortable modification-time column, and a status dot for untracked, unstaged and staged files. Below the tree a git bar shows the branch with ahead/behind counts and expands upward into the 20 most recent commits, with a repository selector when a workspace has independent repositories in subdirectories. A header search box walks the workspace recursively and can open a result or expand the tree at it, preview tabs offer reveal-in-tree and one-click @, the tree is keyboard navigable with ARIA tree semantics, and a right-click entry on the Files tab switches back to the native tree at runtime.

Claude-compatible memory for DeepSeek Harness: a user-chosen memory directory with MEMORY.md index and topic files, the memory tool, and a settings page

Reopens the conversation you were last in after restarting DeepSeek Harness, by keeping a pointer file under the DSH home directory instead of patching engine files.

Cumulative token usage tracking for DeepSeek Harness, with per-model and per-day breakdowns, a settings-page dashboard and an online-changeable data directory.

Discovers workspace Markdown commands, skills and path-scoped rules from .dsh and .agents directories.

Long-term memory for DeepSeek Harness, read-only as of 0.1.0: a scribe_recall tool that reads a plain-markdown memory room, path rules that refuse directory traversal, Unicode smuggling and NTFS alternate data streams, and a bounded index whose truncation is always reported rather than silent.

Appends actionable hints to failed tool results, such as the nearest existing directory when a path does not exist.

Completes @path references in the Web composer: type @ (optionally ./src/ma) to list matching files and directories, pick one with the arrow keys, and the draft receives an @./relative/path reference (quoted when the path has spaces).

A full-page console panel opened from the sidebar, with eight tabs: an overview; a session list read from the sessions service (title, directory, running state, updated time, with filters); installed bundles and Loader entries from pluginInventory and pluginManager, with per-bundle enable switches; a plugin market listing public topic:dsh-plugin repositories from the GitHub search API; a model and provider editor that writes the llm-pi-ai settings namespace through remote.settings.mutate (switch a session model, add or edit providers and their model lists, and store API keys via remote.credentials, merging over the stored profile so fields the form does not expose survive an edit); MCP and skills tabs; and a usage tab that reads the host's loopback /api/dsh-usage/overview document. Services are declared through ctx.inject, so a deployment that omits one mounts without the console rather than crashing.

Approval-free writes into whitelisted directories — wraps the filesystem fence for write/edit tools and auto-answers bash sandbox escalations that touch whitelisted paths, with a settings GUI.

Native DSH right-sidebar tab that lists the ports this machine is listening on together with the process behind each one, and offers per-port actions to open it, copy its address, reveal its working directory or end the process.