Plugins

Zero-touch runtime telemetry for DSH: every session auto-writes task, model tier, tools, failures, duration, status into a human-readable journal with a stats section (pure observer — no tools, no LLM calls, no injection).

模型回合结束后用 LLM 生成 3 条批判性跟进建议，点击即发送 | After a model turn, an LLM generates 3 critical follow-up suggestions shown as click-to-send chips

Shows DeepSeek balance and estimated daily spend, merges external vision-call JSONL records, and adds a per-session dashboard with model statistics, TTFT, cache rate, call logs, cost estimates, filters, and CSV export.

The model switch for DSH: point it at any OpenAI-compatible endpoint, with curated free/cheap DeepSeek provider presets and automatic fallback when a free tier rate-limits you.

Adaptive model routing: per-request complexity classification with automatic provider routing.

Task planning with experience muscle-memory: condition-reflex recall of past solutions, LLM capability matching, and auto-persisted lessons.

Native LLM-provider vision bridge: images pasted in the chat are described by a vision model (Qwen3-VL via pi-ai/llama.cpp) and the text description is fed to text-only DeepSeek for the reply — image admission, routing and compaction all run through harness-native mechanisms, with an LRU description cache and 503 retry.

Tree view of a DSH conversation: editing an old message forks the thread, every version lives in one tree while the sidebar keeps a single entry, branches can be renamed, and a branch can be promoted to a session of its own.

Multi-account Google Gemini provider: picks the account with the most remaining quota, fails over on 429 with exponential cooldown, probes disabled accounts in the background, and ships a Chinese/English settings page.

DeepSeek Harness (DSH) 推理等级自动选择插件：由 Jev System One 模型判断每条消息值多少思考量，按模型声明的等级自动推导档位，上下文信封让「继续」这类追问继承话题深度，低置信度向上取，任何失败都静默沿用原等级。 | Jev-driven reasoning effort per message: per-model ladders derived from what each model advertises, a fixed-size context envelope so follow-ups inherit topic depth, ties break upward, silent fallback on every failure path.

Remote development launcher for DeepSeek Harness: provisions dsh on a remote host via SSH and connects from the local client, with LLM credentials proxied through a reverse tunnel so they never leave the local machine. Ships as both a standalone CLI and a dsh plugin with a session panel, /remote slash command and remote_* agent tools.

TypeSafe Jev (System One decision model) for DeepSeek Harness, end to end: jev_decision (choice/score/noul, parallel, typed answers + confidence), a Settings > Plugins card for the API key and switches, inline tool views showing every decision in the conversation (answer, confidence, latency, cost), an auto-guard (deterministic + Jev risk/loop checks), a local /jev dashboard, and an online verification benchmark (96.3% on 27 labeled questions, median ~283 ms on jev-latest, 2026-09-21). Honest by design: real API only, no mock fallback.

Model switching and reasoning-effort control for DeepSeek Harness: set_model and list_models agent tools let the agent autonomously change its provider, model or thinking depth (reasoning effort) mid-task with a mandatory reason, a provider whitelist and a context-window token-pressure guard, the active model is injected into the runtime context, and Plan Mode auto-transition switches to a configured planning model on entering plan mode and restores the previous execution model on exit.

Runs a thesis-antithesis-seal debate whose verdict is computed from real command exit codes and case counts, never from agent prose.

Multi-language codebase intelligence for MCP-compatible IDEs, the CLI and DeepSeek Harness: chat, search, audit, refactor, and board-ready reports from local code.

One-click Concise output style for DeepSeek Harness: results first, no filler, same work depth. Every final reply opens with a plain-language digest card (select text to copy), scoped per session, working in headless/CLI too, with /concise command, style.md overrides and a local API. Links and file paths inside the digest card render as clickable chips: URLs jump, local files open with the OS default app (v0.9+). A per-step pre-generation contract reminder (v0.11.2+) measurably fixes digest misses on audit-report/table-shaped replies.

Local multi-plan bridge for DeepSeek Harness: use your existing ChatGPT Codex, Claude, ZCode (GLM) and Qwen subscriptions through one loopback proxy, with a settings-page status card and one-click provisioning. No dsh core changes.

Full-screen terminal TUI for DeepSeek Harness with Web-GUI parity, live streaming, agent presets, and DeepSeek blue styling.

L-clone 记忆钩子：DSH 每轮结束(turn/end)自动 capture 到外置大脑；Web GUI 侧边栏「大脑看板」按钮一键打开记忆工作台。

LLM provider adapter that routes model calls through any ACP (Agent Client Protocol) server — Claude Code, Codex, Gemini CLI, Devin and more — with a registry browser, per-server settings UI, lazy authentication, and dynamic model discovery.

人脑式上下文压缩逻辑，大幅减少长任务上下文占用以及token消耗

Irreversible secret-scrubbing guard: rewrites access keys, bearer tokens, and private key blocks into `[REDACTED:<category>]` placeholders before they reach the session log and the model.

Watches agent execution traces and fires an LLM review at every successful turn end when signals trip, crystallizing reusable workflows into registered skills that improve progressively with later runs.

Volcengine Ark Coding Plan LLM provider for DeepSeek Harness: provider-scoped PDF and image input, with Responses API streaming support.