Plugins

Task monitor plus remote chat for DeepSeek Harness. Read sessions, background jobs, subagents, goal rounds, workflows and pending approvals from a phone-friendly page, and talk to any session from it: streamed replies, tool-call progress, stop at any time, switch sessions and models. Also ships a GUI sidebar panel.

Team cost captain: meters every model call (subagents included), attributes each call to a member by workspace rules, prices it with the built-in DeepSeek 2026-09 catalog (peak/off-peak, CNY/USD) or user overrides, and appends it to a local JSONL ledger. Adds hot-reloadable member budgets, 80%/100% webhook alerts, optional over-budget enforcement, a team_cost_report tool with CSV export, and a settings-page team dashboard served over a Typert RPC service.

Per-model system prompt injector for DeepSeek Harness: append custom prompt text to the END of the system prompt for locally configured models, matched by exact provider/model route or a provider-wide provider/* rule (both stack in order). A Settings page lists every configured provider and its models for per-route editing with instant effect, and rules persist to DSH_HOME so they survive restarts. Applies to main sessions, subagents, and workflow children alike.

A visual plugin marketplace for DSH: browse multiple Git sources, install plugins with one click, and wire skills, subagents, MCP connectors, LSP servers and hooks into the harness automatically.

Browser notification plugin for DeepSeek Harness - system toasts when a task finishes, a run errors, or your input is needed (question / approval / plan review), click to open the session, per-type toggles, and no false 'done' while the main session waits on its subagents.

Per-provider concurrency gate on the llm/stream waterfall: holds surplus model requests in a FIFO queue inside DSH so main agent, subagents and compaction take turns on a single-slot local server (llama.cpp --parallel 1) instead of timing out; maxConcurrent / maxQueued / queueTimeoutMs.

Runs a parallel code review squad of read-only reviewer subagents (security, correctness, tests, style), each optionally on its own model, and aggregates one severity-grouped report from /review or a code_review tool call.

Stavros RedTeam for DeepSeek Harness (DSH) — authorized red-team persona + scope-guarded pentest tooling (persona, 24 subagents, methodology and 78 zero-dependency guard tools).

Explicit model routing for Main, Subagents, web_search, and image generation, plus a catalog-driven Composer and Plan Review picker with Model, Effort, Context, Fast, and Thinking variants.

Project skill discovery: enumerate .dsh/skills entries with a SKILL.md present. 项目技能枚举（.dsh/skills 只读清单）

Command-triggered red/blue adversarial review for ANY plan (not quant-specific). /plan-adversarial <plan>: arms a gate that requires the agent to derive two independent subagents (red=attack, blue=defend), have them converge on a consensus plan with NO th

Exposes bounded Prometheus metrics for DSH sessions, agents, LLM calls, tools, approvals, subagents, and jobs, with a Grafana dashboard and a loopback-only endpoint by default.

Minimal prompt x Web PTC capabilities: one-line persona, run_code SDK, subagents, Ralph, web search/fetch, and Windows Git Bash; the general-purpose workflow tool stays hidden.

Active wait for background subagents in DeepSeek Harness: registers a wait_subagent model tool that blocks until a named background continuable subagent settles, then returns its stop reason and closing message — optional timeout, membership gate rejects unknown ids, event-driven with zero polling. Closes the gap between run_in_background fire-and-return and the async settlement notice; wait for a background subagent without polling list_agents.

Memory bridge between DeepSeek Harness and ai-memory: runs ai-memory's finalize-session when a DSH session ends, so the observations it already captured become searchable wiki pages. Skips subagents, queues concurrent session ends, and sweeps sessions abandoned by a forced restart. Requires a separately installed ai-memory server with its DSH lifecycle hooks wired by hand — ai-memory’s own install-hooks does not support DSH; this plugin only finalizes what capture has already recorded.

OpenRouter provider that reads the live /models list, so new models appear without a harness release: models sorted by name with input/output prices, per-model reasoning levels, image input, reasoning_details replay, Anthropic prompt caching, and actual spend per session including subagents from OpenRouter usage.cost, shown under the Web UI composer and by /cost.

Ships a skill about handing work to subagents instead of doing it inline - which tasks are worth fanning out, how many to run at once, how to send them in parallel, and how to check what they returned instead of trusting the summary.

Flip the DSH page over to a glass task board showing live jobs, subagents and session state, with an infinite canvas that unrolls a card into draggable controls, embedded web-app frames, and a conversation box that edits the canvas.

In-process DeepSeek Harness subagent provider that seeds a child with only the last N completed parent turns (fork-lite), leaving official fork untouched.

Unified management and invocation of external Agent CLIs in DSH, fully isolated from the user's native installs; preset Provider, model, reasoning effort, and permissions per CLI, callable directly or like native subagents.

Live token cost for every chat in the DSH Web UI: the chat, its subagents and the whole session tree, priced from a bundled models.dev snapshot of seven providers and 139 priced models, with cache read and write rates wherever the provider publishes them and DeepSeek from its own peak and off-peak table. Every finished answer carries its own price in the transcript, and a chat that is not currently open is priced from the log instead of disappearing. Spend is costed per interval from an append-only JSONL ledger in the project folder, so logged work keeps the tariff it was billed at. On top of it: budget-aware work plans packed against a money limit, routing scenarios that show which models are worth connecting, a generated plan document in English, Chinese or Russian, and a warning when spend belongs to no plan unit.

Keeps every model request inside a per-provider allow-list, enforced at the agent/request waterfall: disallowed models are redirected to a fallback or refused; subagents stop inheriting stale creation-time routes; warns when consecutive requests would make a --models-max 1 llama.cpp router reload models. A one-entry list is a global single-model mode.

Explains which layer killed a failed turn: times every model request (first byte, longest silence, total) and combines that with the error to name the culprit (dsh idle watchdog vs Node undici timers vs SDK timer vs server crash, context overflow, gate queue, tool timeouts, failed compactions) and the exact setting to change; verdicts in the DSH terminal plus a /why command, subagents included.

Task Runner mode for DeepSeek Harness: the main agent decomposes large tasks into independent subtasks, dispatches them to subagents (default local model) in fresh short contexts, and synthesizes results; concurrency and worker/orchestrator models are configurable in the settings panel.