Plugins

🌐 A DeepSeek Harness plugin: pick the language beside the model selector and every reply — sessions, subagents, teammates, side chats — follows. Auto releases the pin.

Live side chat for DeepSeek Harness: an empty read-only conversation in the right sidebar that follows the current session live through a bounded real-time digest, so its progress never goes stale; destroyed when closed.

Seal proven command chains into reusable executable tool scripts for DSH agents. DSH 代码脚本固存，留住模型成功自实现的工作流。

Zero-credential Outlook automation for DSH Harness on Windows: the AI agent drives your LOCAL desktop Outlook client via COM: search/read mail, drafts, reply/forward, calendar, address book, rooms, free/busy. No Graph API app registration, no SMTP passwor

Run Google Antigravity CLI (agy) as DeepSeek Harness subagents, with Gemini web Deep Research and parallel/team modes.

HTTP API for DeepSeek Harness: agents / sessions / event timeline / SSE streams / steering / messages / runs / files / questions / subagents, with archived run manifests, per-key metering and tenant isolation.

Pixel zen-garden plugin for DeepSeek Harness: eight pots and six plants, watering and bloom coins, live Open-Meteo weather, a local helper, and garden_status/garden_action agent tools.

Manage a LAN or Tailscale reachable dsh web instance from inside DSH. dsh web listens on loopback by default; this adds a Settings > Web access panel that detects the machine's Tailscale and LAN addresses, starts dsh web bound to the one you pick, and shows the token URL to open on a phone or another machine. Also stops and restarts it, tails the child log, persists the bind address and port, and registers two agent tools (dsh_web_control, dsh_web_config). Zero runtime dependencies. Anyone who can reach the bound address and holds the token URL gets your dsh's full privileges (files and commands); LAN mode is plain HTTP, Tailscale is encrypted by the tunnel, and the token is single-use and changes on every start.

Multi-machine job board for DSH: 14 board tools, 4 shell tools, a GUI panel, and a dispatcher that hands queued tasks to remote hosts through dsh-remote-agent. Pick the target machine and model from the panel or the board_options tool; rename machines with display aliases while task ownership stays keyed to the host identity (machine). Failed tasks can be retried with their original goal/workspace/model, and the panel checks npm for a newer plugin version. Single-machine use starts its own loopback board; multi-machine use needs the bundled Python board server (tools/board_server.py, started with GP_BOARD_TOKEN set, binding 0.0.0.0 by default) and dsh-remote-agent hosts. The sh_* tools only reach directories on the allowedRoots list.

Peer sessions between DeepSeek Harness instances on a LAN: UDP discovery, an explicit one-way allow list with open or confirm approval, HMAC-signed HTTP with a ±60 s window and nonce replay protection, a `跨端 · <peer>` session for delivered prompts, two-phase file transfer up to 100 MiB with SHA-256 receipts, and seven agent tools that call the named tools listed in `exposedTools` on the receiving side. Its local /cross-agent panel routes are guarded only by a loopback check, without DSH's Host/Origin admission.

SSH workbench for launching remote agent CLIs with durable sessions, terminal, file editing, Git worktrees and local port forwarding. It drives the local OpenSSH client (macOS/Linux; remote Linux with GNU coreutils, tmux optional) to run commands, edit files and forward ports on any host from ~/.ssh/config, with host-key checking defaulting to accept-new; the last 512K characters of each terminal's output are saved under ~/.dsh/remote-agent/terminals. It uses the same package name and mount id as zbzbhs/dsh-remote-agent, so the two collide if installed together.

Security guard for DeepSeek Harness (Cordis): reviews every tool call in pre-execute against SlowMist agent-security rules, denying or flagging risky commands, MCP servers and skill loads, external URLs and repositories, untrusted documents, and EVM or Solana addresses; scans tool results for prompt injection after execution and ships a slowmist_scan audit tool plus an injected security policy prompt. Rules follow the open-source SlowMist Agent Security Skill (MIT).

Read-only diagnostics for a DSH install — boot-crash logs and the cause named in them, config drift against a recorded fingerprint, backup freshness, workspace growth, plugin updates, host-version compatibility, duplicate client ids, and model availability — plus an action layer that always shows a plan first and needs a single-use fingerprint and explicit confirmation before anything is written. A checkup also runs 45 seconds after every host start; when the 3-hour probe cache is stale it sends a one-line test prompt to the default model and a few models of each configured provider (model.probeOnReport can switch this off). Registers the agent tools dsh_steward_report and dsh_steward_act (clean caches via dsh-disk-cleaner, backup via dsh-config-manager, disable or enable a bundle, set the default model; execute takes the plan's hash plus confirm: true) and a system-prompt section telling the agent to run the report first when something crashed.

Adds budgets to agent waiting: caps job_output waits, refuses runaway foreground commands inside run_code, breaks status-polling loops after four no-progress probes, and explains run_code deadlines and parse errors.

Observation-only audit for DSH Agent Teams that records per-request cache usage and prompt-prefix rewrites to JSONL, with no tool or prompt changes.

Loads Google Antigravity rules from ~/.agents/rules and the workspace's .agents/rules into the system prompt: always-on rules in full, the rest as a catalog the model can load with the antigravity_rule tool, or that activates when a touched file matches a rule's globs. The rules are prompt guidance rather than hard enforcement, and the model can toggle them off with the same tool. Settings are written to .agents/rules/settings.json in the workspace.

Sound cues played by the host operating system, not the browser: three bundled WAV files fire through the machine's own audio (SoundPlayer on Windows, afplay on macOS, paplay/aplay on Linux) when an agent's turn completes, when the context is compacted and when the agent is waiting for your answer; subagent turns can be included, cues fired closer than a configurable interval are skipped, and each bundled sound can be replaced with your own file via plugin config.

Sidebar GitHub panel plus github_* agent tools: list your repositories, inspect issues and pull requests, and create, update or comment on issues.

Team presets for AgentTeams: ships audit6 / impl5 / lite2 rosters with seed task DAGs and a per-role report contract, plus a captain doctrine section, so a fresh session opens the right team in one approval instead of a pasted prompt. Needs the engine @nanmicoder/dsh-agent-teams added to the same profile first; the team patch targets that engine's entry by id and replaces its whole agent-teams config.

Multi-workspace agent federation with persistent ticket trees, parallel worker sessions, structured evidence reports, independent acceptance checks, retry/stop and restart recovery. Workers get shell and file tools in the ticket's directory, company_verify re-runs the ticket's acceptance commands in a shell on the host, and scope lists are task authorization rather than a sandbox. State is stored in ~/.dsh/ai-company; the optional company_repo_* tools run separately installed aoci/codegraph commands on any path they are given.

Adds a /grill-me slash command that switches the session to plan mode and prompts the agent to ask 3 to 20 structured questions about your goal, each with a recommended option and alternatives, through ask_user_question before it plans.

Agent skills for Obsidian, vendored verbatim from the official repository kepano/obsidian-skills (MIT, Steph Ango). Six skills covering Obsidian Flavored Markdown (wikilinks, embeds, callouts, properties), Bases (.base), JSON Canvas (.canvas), the Obsidian CLI, Defuddle and Knap. Documentation only: no CLI binaries are bundled, so obsidian/defuddle/knap are installed separately by the user.

Health check for a DSH profile: scores skills and agent instruction files, subagent definitions and presets on one shared 18-bucket S–E grade scale, with a sidebar panel, four dsh360_* tools, and a backup with undo for every write.

Pick elements on a real Chrome, Edge or ego lite page inside DeepSeek Harness, comment and screenshot them, then insert the annotation as a chat reference chip — the agent receives the CSS selector, page position, comment and screenshot, and edits the front end you are looking at. It launches Chrome or Edge with its own profile and a random local debugging port and injects an overlay into the pages; for ego lite it attaches to the instance you already have open unless DSH_VISUAL_ANNOTATOR_ADOPT_RUNNING=0.