LLM Mart Basic
@llm-mart · Joined Jun 2026
Apply Chain-of-Verification (CoVe) prompting to improve response accuracy through self-verification. Use when complex questions require fact-checking, technical accuracy, or multi-step reasoning.
TRIGGER when: adding or upgrading any dependency — library, SDK, framework, API, IaC API version (K8s/Terraform/Helm), CRD, or container image. Use BEFORE writing the call. Forces context7/capy lookup instead of guessing.
Use in pre-implementation (idea-to-design) stages to understand spec/requirements and create a correct implementation plan before writing actual code. Turns ideas into a fully-formed PRD/design/specification and implementation-plan. Creates design docs and task lists in docs/feat
Compare two versions of a skill's instructions to detect degradations and complexity increases. TRIGGER when: user asks to diff, compare, or review changes to a skill's markdown instructions. Asymmetric check — only regressions count. Additions, clarifications, and strengthenings
After implementing a new feature or fixing a bug, make sure to document the changes. Use when writing documentation, after finishing the implementation phase for a feature or a bug-fix.
Compare and merge two design docs for the same feature into a single source of truth. Use when you have competing or complementary design/implementation docs (e.g. from separate design runs) that need reconciling into one unified document.
Produce or update a domain-reference kit — the durable glossary + divergences/traps pages for a bounded context. Use for domain modelling, information architecture, building a domain glossary or domain model, mapping domain concepts to code (bindings), or surfacing the decidable
Designs, reviews, revises, and templates deployable system prompts for agent projects. Use whenever the user asks to write, improve, review, or debug an agent's system prompt, system message, or role/boundary instructions — including Chinese requests such as 系统提示词、角色设定、角色和边界规范、提示
Делает одностраничный сводный снимок бизнеса для собственника — позиция по деньгам (бухгалтерия), тренд продаж (платежи), движение по воронке (CRM), обязательства недели (календарь), срочный список наблюдения (почта, мессенджер) и единственное самое важное на сегодня. Сам пробует
Ранжирует топ-5 лидов, которым стоит позвонить сегодня, готовит тезисы разговора из истории переписки, бронирует время в календаре и набрасывает follow-up сообщения. Принимает необязательные аргументы количества и даты. Командная форма: вызывается по имени.
Берёт согласованный контент-бриф и выполняет кампанию от начала до конца: строит календарь публикаций, генерирует макеты для постов в соцсетях через ~~дизайн (VistaCreate, Supa, Поликрафт), пишет тексты подписей и писем, ставит публикации в очередь через ~~crm. Макеты делаются то
Берёт дебиторку и кредиторку, историю сроков поступления денег и известные постоянные расходы из бухгалтерии (1С, МойСклад) и платёжных сервисов (ЮKassa, Тинькофф, банк-выписка) — или из выгрузки CSV — и строит прогноз денежного потока на 30/60/90 дней с доверительными коридорами
Закрывает месяц — сверяет данные бухгалтерии с эквайрингом и выписками, помечает расхождения, пишет ОПиУ-сводку, выгружает пакет закрытия. Принимает необязательные аргументы месяца и места сохранения. Командная форма: вызывается по имени.
Анализирует данные о продажах из ~~платежи и ~~бухгалтерии (ЮKassa, 1С, МойСклад), находит хиты и зависшие позиции, накладывает сезонность и выдаёт приоритизированный контент-бриф на 30 дней: что продвигать, какие акции запускать, что придержать. На выходе только стратегия — без
Лёгкая проверка договоров (поставка, услуги, подряд, NDA, SaaS) по праву РФ для малого бизнеса без штатного юриста. Читает договор из локального файла, вложения в почте (Яндекс 360) или ЭДО (Диадок/СБИС); подсвечивает рискованные условия; объясняет риски простым языком; отдаёт ра
Проверка контрагента (юрлица/ИП) по ИНН перед сделкой, отгрузкой в долг или предоплатой. Двухскоростная: быстрый quick-scan по deal-killer-сигналам (ликвидация/банкротство/недостоверность ЕГРЮЛ/дисквалификация/крупные долги), затем полное досье по запросу. Собирает открытые данны
Сканирует CRM на зависшие сделки, дубли контактов и незаполненные поля, затем чинит то, что собственник одобрил. Принимает необязательный аргумент области: сделки, контакты или всё. Командная форма: вызывается по имени.
Держит CRM в актуальном состоянии без того, чтобы собственник её открывал: создаёт и обновляет контакты и сделки из контекста переписки и календаря, заносит примечания и звонки, помечает зависшие записи. Скилл «хватит вбивать данные руками». Используй, когда собственник просит об
Движок круговой сверки: берёт результаты из НЕСКОЛЬКИХ источников (агрегаторы, госреестры, веб, выгрузки) и сводит в один ответ с дедупликацией, оценкой уверенности (свежесть × авторитет × согласие источников), tier-маркерами и явным показом расхождений. Не выбирает молча одну ве
Собирает споры по платежам, обращения и отзывы клиентов из CRM, почты и поддержки (плюс вставленные или выгруженные отзывы с Яндекс Карт, 2ГИС, Авито, VK, Отзовика) в отчёт по темам с дословными цитатами и списком «сделайте эти три вещи на этой неделе». Триггеры: «что говорят кли
Fourteen posts of being wrong in production, compressed to checkboxes
Healthy nodes, a quiet network, 300 restarts in three days, and a latency budget measured in milliseconds
Discovery worked. Ping worked. Every TCP connection timed out, and later the tunnel only worked when someone had a terminal open.
Every VM came back. The cluster did not. Declarative systems converge on config, and the datapath isn't config.
A surprising share of AI-in-the-terminal failures aren't the AI. They're zsh, and a version of bash from 2006.
A Claude Code plugin turns standalone project configuration into a namespaced, installable extension that teams and communities can update as one unit.
None of the safety came from the model. It came from six boring habits.
Skills package instructions and references. Subagents run work in a separate context and return results. They solve different problems and can be composed deliberately.
Six hours in, one step left, everything green, and the incident that didn't happen
CLAUDE.md carries persistent project context. Skills load reusable procedures when relevant. Separating stable facts from task-specific workflows keeps both easier to maintain.
Twenty minutes recovering secrets that never existed, and the one sentence from a human that ended it
An API request routing a model's tool call through an approval gate to a remote MCP server
31 config keys, two audits, and why the first one was wrong in both directions
The official MCP Registry stores standardized server metadata rather than package code. Publishers verify a namespace, describe installation or remote access, and submit immutable versions.
Everyone looks at the Dockerfile. The file that actually leaked the key was the project file.
Remote MCP authorization uses established OAuth standards, but secure integration still requires issuer validation, least-privilege scopes, protected token handling, and server-side enforcement.
"Copy it over and switch the reference" is two steps, and the outage lives in the one nobody checks
stdio fits local processes and prototypes. Streamable HTTP fits hosted services and shared integrations. The right choice follows where the capability runs and who must reach it.
The most important rule wasn't about what I could change. It was about what I was allowed to display.
Tools perform operations, resources expose readable context, and prompts provide reusable templates. Choosing the correct primitive makes an MCP server easier to understand and govern.
/shard
Shard
`crabbox shard` forks a [checkpoint](checkpoint.md) into `--count` leases
/share
Share
`crabbox share` grants other people access to an existing lease through the
/ssh
Ssh
`crabbox ssh` resolves a lease and prints a ready-to-run `ssh` command for it.
/status
Status
`crabbox status` prints the current state of a lease: its slug, provider,
/stop
Stop
`crabbox stop` ends a single lease. For coordinator-backed and direct cloud
/sync-plan
Sync plan
`crabbox sync-plan` prints the local sync manifest and its size hotspots
/tunnel
Tunnel
`crabbox tunnel` opens one foreground SSH local port-forward to a resolved
/unshare
Unshare
`crabbox unshare` removes sharing rules from a coordinator-backed lease. It is
/usage
Usage
`crabbox usage` reports lease cost and usage estimates from the broker, broken down by user, organization, or the whole fleet.
/verify
Verify
`crabbox verify` checks a signed run receipt produced by `crabbox run --attest`.
/vnc
Vnc
`crabbox vnc` prints (or opens) the connection details for a desktop-capable
/warmup
Warmup
`crabbox warmup` leases a box and waits until it is ready: it provisions (or
/watch
Watch
`crabbox watch` runs a command on a warm lease, then watches the local
/webvnc
Webvnc
Linux portal viewers default to **Match window**: only the connected controller
/whoami
Whoami
`crabbox whoami` calls the broker's `/v1/whoami` endpoint and prints the
/add-dep
Add dep
Vet a new or changed third-party dependency for license, provenance, and supply-chain risk before any install runs.
/adr-status
Adr status
Inspect ADR health read-only; optionally select one ADR with --adr N.
/adr
Adr
{{SKILL_ENTRY:decision-lifecycle}}
/audit
Audit
Assemble the governance record for a range — commits, overrides, ADRs, sprint auto-decisions, open questions, checkpoint findings — into one dated audit packet. Read-only.
/btw
Btw
Lightweight Q&A about the project — answer from context and return, no routing, no state change.
Implementation of Podlite markup language
12 views 0 likesOpen-source AI video workbench. Bring any model or your local ComfyUI, and let Claude Code / Codex / Cursor direct it over MCP — storyboard, references, generat…
15 views 0 likes《深入理解 AI Agent:设计原理与工程实践》(李博杰 著)开源主仓库:全书正文、编译版 PDF 与按章配套代码
8 views 0 likesDistilly — Distill how they think into reusable Skills for any Agent or Bot. Formerly Colleague Skill(原同事 Skill).
16 views 0 likes本地优先的类飞书 Markdown 编辑器,内置可审改的 AI 助手 | Local-first WYSIWYG Markdown editor with a reviewable AI agent
14 views 0 likesA lightweight all-in-one terminal with 20+ protocols — SSH, RDP, SFTP, databases, Kubernetes and more. With a built-in autonomous AI Agent that plans and runs m…
27 views 0 likes一款专注于桌宠方向的AIAgent。特点是一键安装、功能齐全、极低开销、完全暴露上下文、全功能插件化、AI自主插件开发、永久唯一会话、类游戏引擎交互策略。具有极高的扩展性和拟人程度上限,非常适合想长期培养和自定义需求高的用户。
11 views 0 likesControl Gmail, Google Calendar, Docs, Sheets, Slides, Chat, Forms, Tasks, Search & Drive with AI - Comprehensive Google Workspace MCP Server & CLI Tool
12 views 0 likes一个先接住情绪、再分析关系并给出可执行策略的 Codex 恋爱军师,内置心理、法律、社会、人文、哲学、婚姻家庭与性学知识库,支持多元关系。
11 views 0 likesAI-powered SRE platform — read-only infrastructure diagnostics with deep investigation, security governance, and team collaboration
27 views 0 likesAn open-source AI desktop companion inspired by Cyrene, combining immersive Chat, personalized long-term memory, and an agentic Work mode.
12 views 0 likesA local-first AI agent with persistent memory, emotional intelligence, and a peer-to-peer skills economy.
15 views 0 likesThe open-source, multi-agent alternative to Claude Tag. @ any agent, wherever work happens, your agents work alongside your team and each other, learning as the…
27 views 0 likesSimple LLM service identification - translate IP:Port to Ollama, vLLM, LiteLLM, or 60+ other AI services in seconds
14 views 0 likes🔥 Official Firecrawl MCP Server - Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.
15 views 0 likesClaude in Chrome, reverse-engineered and open-source. No domain blocklist. Any Chromium browser. Same 18 MCP tools, same performance.
14 views 0 likesAgent-native flight & hotel search and booking — MCP server, CLI, and Python/JS SDKs. Hundreds of airlines plus the major booking sites, with per-flight reliabi…
19 views 0 likesOpen-source desktop AI agent for tools, files, knowledge, workflows, and real deliverables.
12 views 0 likesOpen-source AI job search: scan job portals, evaluate listings into a structured A-H report with a global 1-5 score, tailor your CV, track applications — runs l…
25 views 0 likesPrivacy-first Codex skill for discovering, validating, completing, and tracking job applications
14 views 0 likes