Плагины

Плагин предметной области разработки на Rust для DeepSeek Harness (dsh): пирамида верификации fmt→clippy→build+test→miri+audit(+semver) как инструменты для модели + управление на уровне графа (anti-Goodhart: 8 правил, включая плотность #[allow] / удалённые тесты, blind-spot через Cargo.toml и pub

Детерминированный обзор безопасности кода: более 40 правил, энтропийное обнаружение секретов, обзор подготовленных diff, экспорт SARIF, приёмка базовой линии, инвентарь зависимостей SBOM-lite и самопроверка здоровья.

Менеджер правил, команд и навыков для dsh: слэш-команда /rules, панель настроек с визуальным редактированием правил (добавить/изменить/отключить/удалить, поддержка свободных зон), пользовательские команды с аргументами {input}, управление навыками и автоматическое резервное копирование с восстановлением AGENTS.md в один клик.

Наводит порядок в навигации настроек DSH: складывает записи сторонних плагинов под сворачиваемыми строками групп, добавляет пользовательские группы в стиле закладок, переключатель сворачивания и автоматическую классификацию (теги маркетплейса → правила имён → ваша модель ИИ).

Aborts a DSH reasoning stream that starts repeating itself instead of letting the turn stall: five rules over the streamed reasoning text (periodic unit, repeated block, repeated line above a share floor, 64-gram density, and a long trailing run of punctuation or whitespace), plus a JSONL fire journal, a Web GUI log panel, and opt-in recovery that retries with a corrective message.

One hub over every coding IDE you run — dsh, Claude Code, Codex, ZCode, Qoder, CatPaw, Trae, OpenCode. Zero-API: it reads each tool's own on-disk state and gives a session inventory, per-IDE usage with cache-hit rates, one-line resume commands, and a single .hub/ rule body wired into all eight via marker-delimited pointer files. The domestic-IDE adapters (Trae/Qoder/CatPaw) were verified against the installed binaries — a gap no overseas tool covers.

DeepSeek Harness approval guardian: a reviewer model decides permission requests before they reach you, with a System One decision-API backend (TypeSafe Jev, SiliconFlow, self-hosted Laya), a configurable low-confidence policy and an allow-rule whitelist.

Connects one existing DSH session to QQ through OneBot/NapCat without starting a new session, and uses a fail-closed whitelist plus @-mention, nickname and configurable-probability wake rules to decide which messages reach the model.

Reasoning-effort gear presets for third-party models: 16 vendor/model presets with per-mode custom wire values, written into dsh-llm-pi-ai with pre-write validation and per-rule rejection; per-model switching between OpenAI Chat Completions, OpenAI Responses and Anthropic Messages with automatic route regrouping; plus a composer Optimize button and a settings section that rewrite the input text with a model you pick.

Checks each step of a plan before you approve it, using command rules plus Jev or an OpenAI-compatible endpoint with logprobs, and marks steps that can't be undone, affect things outside the workspace, weren't asked for or are too vague on the plan review card.

Apply configurable allow, deny, and approval rules to dsh tool calls, including shell command patterns.

Alibaba Open Code Review for DSH: host-model delegation with native preview/rule tools and a review skill, plus optional OCR-managed reviews.

Audio cues for DeepSeek Harness: a short synthesized sound when an agent needs confirmation, finishes a turn or a background job, or fails, with five sound packs, per-event toggles and volume, and do-not-disturb rules stored in the host settings document.

DSH plugin that registers the ai-eight-honors-eight-shames skill: eight checkable working rules for coding agents (verify before calling, confirm ambiguity, cite evidence, reuse before creating, test before claiming done, follow local conventions, separate fact from inference, keep changes reversible) plus a ninth rule that logs each user request and the change made in two fixed report files at the workspace root.

Ponytail lazy-senior-dev ruleset for DeepSeek Harness: a 7-rung minimal-code ladder (YAGNI, reuse, stdlib, native, installed deps, one line, minimum), six skills (main rules, over-engineering review, repo audit, debt ledger, gain scoreboard, help), a /ponytail session command for lite/full/ultra/off levels, and optional always-on ruleset injection via config. Vendored verbatim from upstream ponytail (MIT), zero tools registered, clean uninstall.

Cross-session experience memory for DSH. Lessons are evidence-graded, injected when the turn is about the same thing, and shown just before an action only when the record declares which call it applies to. A rule that holds whatever the turn is about — "always answer in Chinese" — can be marked **standing** and is then carried every turn: measured on 24 isolated real-model turns, a standing rule is followed 6/6 against 0/18 for the controls (empty store, same record unmarked, unrelated standing record). A guess still cannot become always-on: a standing record needs a verified passage like any other. What nobody looks up is retired, reversibly. Five tools, seven slash commands, zero runtime dependencies.

Pre-execution safety valve for DSH that judges with TypeSafe Jev: it hooks tools/pre-execute and decides every bash/pwsh call by first applying offline static rules, then by asking the Jev model — the System One model from TypeSafe, which returns a structured decision instead of prose — one yes-no question ("will this command irreversibly delete or overwrite real data?"), turning the answer into allow, revise, block or escalate; revise hands the model a safer rewrite, escalate offers a one-shot human token; exhausted credit or a missing API key degrades loudly instead of failing silent, and a missing key is requested in the conversation itself (`guard key set`, read from stdin); every verdict is appended to a shared audit log and the text people read is bilingual zh-CN/en.

BaziDIY ontology plugin: deterministic bazi/wuxing/bead/style rules as DSH tools (calculate_bazi, propose_designs, generate_design, save_design, load_design), plus bracelet SVG rendering and a bead editor; no LLM calls inside execute.

Ships a skill about writing DSH skills - SKILL.md frontmatter and root precedence, how to phrase a description so it actually triggers, length and splitting rules, and an A/B check that the finished skill changes agent behaviour. The package also carries the checker the skill uses to validate a skill directory.

Reads .mcp.json from the .dsh home and from the agent workspace, mounts every mcpServers entry into DSH, and adds an MCP tab in the Web UI for listing, adding and unloading servers.

Injects the rule files under $DSH_HOME/rules and <cwd>/.dsh/rules into the conversation at the first agent pre-step, re-injecting them after compaction.

Windows system-drive (C:) cleanup for DSH: rule-based scan, five-tier safety grading, a staging area with a rollback ledger, UAC elevation, and migration to another drive — a model tool plus a Web GUI panel.

Language-pack framework that rewrites model replies: A is an independent LLM rewrite, B injects FTS5 terminology retrieval into that rewrite, C applies local dictionary and regex rules, and composite runs a pipeline such as B then A then C. Ships a GitHub pack marketplace, a ZIP installer with path, size and file-type checks, and a mandatory risk disclaimer gate.

Task board plugin for the DSH web GUI: a sidebar entry plus a multi-column kanban view with real execution through DSH sessions, schedules, session rules and auto-cruise automation. Board truth lives on the host and syncs to every device over SSE (the bro