プラグイン

It logs how you talk to your AI — the 7th "final version" gets a bold red page, a late-night rewrite request gets underlined, a "thanks" earns a gold star, and a post-AGI verdict decides your fate. All local, rule-based, bilingual.

Embeds a remote DeepSeek Harness GUI in the local DSH window over an SSH tunnel, lists the remote sessions, and switches between the local instance and any remote host.

Connects DeepSeek Harness to the hosted Local Commander MCP server at localcmd.app via mcp-remote with browser sign-in, to read files and Git, run the checks you approve and check local services on your Windows computers. Requires a Local Commander account and the paired agent running on the Windows PC.

Encrypted local credential vault for DeepSeek Harness: sidebar record management, an info_records tool for pre-sign-in lookups, opt-in chat capture, per-record manual-only mode, and AES-256-GCM storage protected by Windows DPAPI, the macOS Keychain, or the Linux Secret Service.

Rewrites the model-visible copy of system, developer, user, assistant and tool messages into stable HMAC placeholders before a request leaves the machine, while local session logs, the chat transcript and restored tool calls keep real values.

Outbound route pool for DeepSeek Harness. Ships 15 seed routes plus a 20-entry candidate pool covering 7 target kinds (GitHub raw/codeload/main, npm registry, PyPI index, HuggingFace, general). Routes are verified before use: upstream IP lists go through a TLS handshake with full certificate-chain validation (a bare TCP check can be answered locally and report dead IPs as healthy), and failures advance a cooldown ladder of 2/5/15 minutes with the fastest healthy route picked per target kind. Routes are applied by rewriting the URL with one of four rules (prefix, registry, index, host-replace), reaching git, pip, npm, pnpm, gh, curl and wget through three paths: the shell seam (download commands the model runs through the bash tool are intercepted and rewritten automatically; commands wrapped by the pwsh tool, and fetches made by the host itself or the renderer, are not covered), two registered tools the model can call directly (line_pool_status, line_pool_run), and child-process spawn. Multi-line commands keep their line breaks and only the download segment is touched. URLs carrying credentials (user:token@host) are never rewritten, so tokens are not forwarded to third-party mirrors. No global config is modified. A sidebar panel lists every route with latency, failure reason and cooldown countdown, supports per-route retest and enable/disable, and shows live TUN/proxy status from network-adapter and loopback port checks. The candidate pool is re-verified every 24 hours so newly usable routes are added and dead ones dropped. 570 tests, 87% line coverage.

Training Lens opens a local TensorBoard scalar dashboard with selectable metrics, explicit improvement direction, optional EMA smoothing and Chinese/English UI; requires an existing Python environment with TensorBoard.

Keep a DSH profile identical across machines through any WebDAV server. The plugin collects the five parts that make up a setup — models (the llm-pi-ai line of cordis.patch.yml), defaults (the agent-default-model line), patch (the remaining loader lines), plugins (package.json dependencies plus dsh.profile.bundles) and dirs (extra absolute-path directories picked in the panel, carried whole and restored to the same path, add-only, at most 5 with 2000 files / 8 MiB each) — encrypts them on this machine with a passphrase (AES-256-GCM, PBKDF2-SHA256 at 600000 iterations) and uploads one ciphertext file to <address>/dsh-webdav/dsh-profile.prop, so the server only ever holds the sealed copy (the older dsh-profile.dshsync name is still read). Merging is per part and three-way against the last synced baseline; a conflict keeps the newer copy and archives the losing one under sync/ (latest 5 per tag), every overwrite is sealed beforehand, and deletions propagate except in the plugins part, on the first sync, and in the dirs part, where dropping a directory on one machine only stops syncing it there and never deletes its files. Uploads carry ETag If-Match and If-None-Match so a concurrent change is refused instead of being clobbered. Sync runs on demand or on a 3/15/30/60 minute timer from a settings panel, and an optional local forwarder (off by default, loopback-only) re-serves the same target with locally generated credentials, rewritten PROPFIND hrefs and .prop aliases for extensionless resources. The local state lives in <profileDir>/.dsh-webdav-sync/ (sync.json is written 0600 and holds the WebDAV password and the passphrase).

Inspects ONNX models (inputs/outputs, shapes, opset, operators, size) and benchmarks them with ONNX Runtime on the local machine or on a remote target over SSH, reporting average/P50/P95/P99 latency, FPS and the execution provider actually used, with PASS/FAIL gates for FPS, P95 latency and model size. Requires Python 3.10+ with numpy, onnx, onnxruntime, and psutil installed (pip install numpy onnx onnxruntime psutil); remote targets need python3, numpy, and onnxruntime.

Moves a DSH profile to another machine with its plugins, not only its files: packs the plugin bundle selection and the whole settings layer into one JSON file, then on restore writes the configuration back, restores local plugin sources and plugin-owned data, and installs the plugin bundles the backup recorded that the new machine is missing (plugin manager first, pnpm as a fallback).

Speaks a cue for DeepSeek Harness on Windows: a configurable phrase when an agent turn completes normally, a second phrase when the agent needs you to choose — a question, a plan approval, or a permission request — and a time-of-day greeting when a new conversation starts. A small bubble in the lower-left corner mutes and unmutes every cue with one click, with its own toggle chime. Phrases are synthesized with a Microsoft Edge neural voice (skipping aborted and errored turns, subagent turns, and rounds that a goal is still driving), cached under the DSH home directory and replayed locally, with the desktop SAPI voice as the offline fallback.

Turn-finished music reminder: when an agent turn ends, play a ringtone or any local audio file (bundled synthesised telephone ring, custom absolute path, or an in-page upload), with a floating chip to pause, resume or dismiss it and loop-until-paused playback; configured in Settings, defaults overridable per deployment.

Bidirectional GitHub repository sync for DSH: git sync with conflict rollback and pre-push checks, Actions/CI monitoring with per-step run detail, repository add/remove/rename, a web panel with local and remote file browsers, scheduled automation, and six agent tools.

Redundant branch cleaner: finds branches whose work already landed in the mainline — merged, squash-merged, patch-equivalent after rebase or cherry-pick, upstream-gone, or PR-merged — and deletes them when asked. It only reports by default, never switches branches or rebases, and backs every deleted local branch up under refs/backup/branch-cleaner/ (deleting remote branches has no backup).

Registers native doubao_generate_image and doubao_image_status Agent tools to automate Doubao Web (doubao.com) via Tabbit Browser, automatically extracting watermark-free 2.7K wm0 original images into local files and multimodal chat attachments.

Self-maintained DSH web task board with a human acceptance gate, card kinds and editable notes on executed cards.

Evidence-first reuse skill for DeepSeek Harness: before implementing, porting, choosing dependencies, or troubleshooting a tool, it searches the local project and machine for existing solutions in layers, judges reuse on checkable evidence, refuses reuse when essential requirements or risks are unmet, and reports what could not be verified. Ships as a DSH bundle that registers the skill on ctx.skills.

Bundles the kb-curator skill, which keeps a local three-layer knowledge base and refuses to index anything the user has not approved. The L2 retrieval layer needs @kenz1117/dsh-engram installed separately, and the engine runs on Windows only.

Reads and writes WPS 365 cloud documents (smart documents and .docx/.xlsx/.pptx) through the locally installed wps365-cli, with conflict checks before writing; requires a WPS 365 enterprise account.

GPT-SoVITS voice output for DSH: a speak tool and /voice command, a read-aloud speaker button under each reply, and a settings tab to switch voices, tune synthesis, manage the local service and one-click install a voice pack. Windows only; needs a local GPT-SoVITS checkout with trained weights and its Python environment.

Meeting notes for dsh: one meeting per conversation, transcribed and speaker-separated on the machine itself, with points, assignments, handwritten notes, reviews and monthly scores in the same panel. Exports SRT/Markdown/Word and packs a meeting into a portable zip another dsh can import; confirmed assignments go to DingTalk/WeCom in one click, following your robot settings. Recognition models are swappable (a stronger local model can be imported from a file), and cloud recognition is off by default with no request made before consent. Word export needs Python and python-docx; the first run downloads a recognition model (28 or 77 MB).

Local offline prosody analysis for DeepSeek Harness: reports how something was said — speed, pitch, pauses, energy — alongside the transcript.

Registers two tools for DeepSeek Harness. `doc_ocr` turns scanned PDFs and images into structured text (per-page text, financial tables with a balance-sheet tie-out check, contract-clause hits) through a local RapidOCR engine or a Zhipu GLM-OCR cloud engine, and `doc_preprocess` turns a directory of raw files into an Obsidian-ready knowledge base through a seven-stage pipeline with backend routing and sha256-based incremental skipping. Numbers that fail the tie-out are reported, not invented. The local engine needs Python 3 with rapidocr-onnxruntime, pypdfium2 and Pillow (pip install -r requirements.txt); the GLM engine needs ZHIPU_API_KEY.

Reads a local video or page link with ffmpeg and Paraformer ASR, returning a timestamped transcript plus key-frame images the agent can open; finds candidate videos by keyword and installs its own runtime dependencies on first use. Requires Python 3.8+ on PATH; video_discover also needs a one-time douyin/xiaohongshu login.