AI & ML

awaik Bundle

Диагностировать подключение LidFly MCP v3 в Claude Code, Claude Desktop, Codex, Cursor, OpenCode и VS Code: неверный config, timeout транспорта, незавершённый MCP OAuth, конфликт статического Authorization с OAuth, отсутствие provider-подключения и устаревшие skills. Использовать при Failed, Authenticate/Login, Authentication failed, SSE 405 и недоступных инструментах.

awaik Bundle

Безопасно восстанавливать transport errors и эскалировать проблемы LidFly MCP через read-only support_prepare_report и подтверждённый support_send_message. Использовать при unexpected/internal support_hint, повторном timeout read-вызова после retry и connectivity probe или неизвестной отсутствующей возможности после широкого search_tools; известные ограничения API провайдера не эскалировать.

codyswanngt Skill

Health-check the LLM Wiki: orphan pages, contradictions, stale claims, broken internal links, missing index/log coverage, structure violations, and secret/tenant leaks. Read-only — reports findings, does not fix them.

codyswanngt Skill

Answer a question from the LLM Wiki with citations. Reads the index, drills into relevant pages, and synthesizes a cited answer. Read-only by default; files new synthesis back only when explicitly asked.

codyswanngt Skill

Scaffold, repair, verify, or upgrade the project's LLM Wiki from its config. Asks the wiki's purpose and README mode, renders the contract snapshot, scaffolds the canonical folders, and seeds the staff roster. Idempotent and non-destructive.

codyswanngt Skill

Onboard a user to the project via its LLM Wiki: interview them about themselves in relation to the project, then give a guided tour and sample questions. Read-mostly by default (session-local); --save-memory persists the capture to project-scoped memory only. No PRs, no PII written into the wiki, never global memory.

codyswanngt Skill

Scaffold, repair, verify, or upgrade the project's LLM Wiki from its config. Alias for lisa-wiki setup in the Lisa setup command family.

codyswanngt Skill

Connect staff roles to Slack via OpenClaw using a facilitator/specialist hub-and-spoke model — register the app, create/reuse the facilitator channel, wire routes, validate, and run an end-to-end route test. Requires /lisa:setup-openclaw first.

codyswanngt Skill

Connect staff roles to Telegram via OpenClaw using a facilitator/specialist hub-and-spoke model — register bots, create/reuse the facilitator topic, wire routes, validate, and run an end-to-end route test. Requires /lisa:setup-openclaw first.

bioconductor Skill

Help create a new AI agent skill through collaborative Q&A

bioconductor Skill

Validate that a skill conforms to the ai-agent-skills repository standards

bioconductor Skill

Verify that bioconductor/ai-agent-skills are available and help users find the right skill

chakhdz Skill

Use when backing up this account’s skills, routines, and agent notes. First a very simple plan, then wait for yes. Never delete. Copy only; keep SKILL.es.md.

mine-fnl Skill

Use when authoring, reviewing, or refactoring a SKILL.md file that follows the open SKILL.md spec at agentskills.io. Applies when the user asks for "write a skill", "review my SKILL.md", or "what should go in the frontmatter". Do not use for authoring plain Markdown documentation, for writing prompt templates, or for designing product features — SKILL.md is one specific file format with strict rules.

rtfm-it-services-llc Skill

Use this skill whenever measuring or judging your MSP's business health or a client relationship: MRR, margin or profitability per client, effective hourly rate, "are we making money on [client]", tickets per user, SLA attainment, churn, aged receivables, revenue concentration, the monthly owner review, the underlying numbers behind the client health scorecard (msp-qbr owns the client-facing scorecard document), or any version of "should we fire this client" or "this client is a problem". Also trigger when building dashboards, review sheets, or reports about the business. Apply alongside msp-pricing (the cost model and floor that give the numbers meaning), msp-helpdesk (ticket hygiene is the data source), msp-qbr (where client-facing consequences land), msp-client-comms (the repricing letter), msp-legal (waivers and exits), and msp-offboarding (the exit path).

richiehakim Skill

Executive role for an agent that dispatches subagents.

richiehakim Bundle

Guide for controlling Adobe Illustrator via MCP tools and osascript.

richiehakim Bundle

Agent roles, models, and the dispatch pattern for delegating to subagents.

richiehakim Skill

Tool for preparing PDF figure files for embedding in LaTeX (e.g., with \includegraphics).

richiehakim Bundle

Workspace layout, artifact storage, and behavior rules shared by every agent (managers and subagents alike).

surgepix Bundle

Edit images from reference image(s) and a natural-language prompt via SurgePix. Do NOT use when the user wants 超现实主义风格 / 超现实主义 / surrealism / 改成超现实 (use surgepix-generate-restaged-cinematic) or 极简诗意风格 / 诗意风格 / 改成极简诗意 (use surgepix-generate-photo-poetic-editorial). Use when the user says "图片编辑", "改图", "重绘", "image edit", "edit this image", "把图里的…改成…", or wants general image editing from references — but "改成" plus 超现实/诗意 is a locked-style restyle, not this skill. Language: write --prompt in the user's conversation language. Output rule: show ONLY the download URL from script stdout verbatim in a code block; NEVER fabricate or retype URLs. Local reference paths are uploaded first — API accepts URLs only.

surgepix Bundle

Generate 16:9 horizontal article illustrations (1536×864, hand-drawn style) via SurgePix for blogs, WeChat Official Account articles, and tweets. Use when the user wants 文章配图, 公众号配图, 博客插图, 推文配图, or horizontal editorial illustrations. Language: match user's conversation language for on-image text (English prompt → English labels; 中文 → 中文). Output rule: API returns only ONE download URL — single image = image URL; multiple images = ZIP URL (resultType zip). When ZIP, show ONLY that download link; NEVER fabricate per-image URLs. Do NOT use for 小红书/RED/竖版套图 — use surgepix-generate-xhs. If the user only says 配图 without platform, ask first.

surgepix Bundle

Restyle 1–5 reference photos into controlled surreal cinematic stage illustrations via SurgePix (server-locked style). Prefer this over surgepix-image-edit when the user says 超现实主义风格, 超现实主义, surrealism, 改成超现实, 帮我把这张图改成超现实主义风格, 超现实重构电影化插图, 受控超现实舞台, 前卫电影感, 抽象几何色场, 大型色块构成, or "restaged cinematic". Language: write --prompt in the user's conversation language. Output rule: show ONLY the download URL from script stdout verbatim in a code block; NEVER fabricate or retype URLs. Local reference paths are uploaded first — API accepts URLs only. Do NOT use for general image edit (surgepix-image-edit), poetic editorial (surgepix-generate-photo-poetic-editorial), or topic-based article illustrations (surgepix-generate-illustrations). If the user only says 把照片做成插画 without naming 超现实/诗意, ask first. 「超现实主义风格」is an explicit style — call this skill, do not ask and do not use image-edit.

fhgomes Bundle

Transcribe audio and video into text and subtitles — locally (faster-whisper large-v3 on the GPU) or remotely (youtubetotranscript.com for YouTube videos). Use whenever the user asks to "transcribe", "get the text of this video/audio", "generate subtitles", "SRT", mentions Whisper, or wants the text of a talk, mentoring session, recorded meeting or YouTube video. Also use when comparing transcription quality or when a previous transcript has loops/hallucinations (the same word repeated over and over) — the fix is the large-v3 model. Ships a ready script (scripts/transcribe_gpu.py) — do not rewrite it from scratch. Also triggers on the equivalent phrases in other languages.

fhgomes Bundle

Use when maintaining, configuring, diagnosing, or updating an OpenClaw instance — cron jobs, agents, channels, config changes, version upgrades, service restarts, or troubleshooting delivery/exec/channel issues. Also use when reviewing agent instructions or bootstraps for optimization.

guaardvark Skill

Operate a running Guaardvark: GPU and VRAM state, plugin start/stop, logs, Celery tasks, the Interconnector sync to other machines, overnight RAG autoresearch, and infographics. Use when the user asks what is using the GPU, why a job is stuck, to start or stop ComfyUI/Audio Foundry/Swarm, to sync another box, to run autoresearch, or to make an infographic.

guaardvark Skill

Connect this agent to a running Guaardvark (self-hosted AI studio) and check what it can do right now. Use first when the user mentions Guaardvark, asks to generate images/video/music/voice locally, run a Film Crew, train a LoRA, launch a swarm, or when a guaardvark MCP tool is missing or a call fails with "not reachable" / "plugin offline".

guaardvark Skill

Add any Hugging Face image or video model, checkpoint or LoRA to Guaardvark from a URL, list what is installed, and download registry models on request. Use when the user pastes a huggingface.co link or a .safetensors URL, asks "can it run <model>", or wants a new LoRA or checkpoint available in the Studio.

ice-lfernandes Skill

Reads the execution trail the `audit` hook writes into `.claude/audit-usage/` and consolidates it across runs — spend per skill and agent, duration, failure rate, permissions granted, and which report to open next. Answers "what has this project cost me so far", which no single report answers. Explicit invocation only.

ice-lfernandes Bundle

Decides which Claude Code extension resolves a scenario — auto-invocable skill, manually-invoked skill, subagent, rule in `rules/`, `CLAUDE.md` section, or MCP server connecting to an external system — and writes the files only after approval. Interviews first, proposes scored options with sources, writes afterward. Explicit invocation only.

ktaroabobon Bundle

「〜するスキルある?」「〜を楽にするツール無い?」「どうやって X をやる?」のように、既存のスキルで解ける可能性がある依頼を受けたとき、手元にインストール済みのスキル → このリポジトリ(ktaroabobon/agent-skills) → GitHub / skills.sh のエコシステムの順に探し、提供元と中身を確認してから提案する。「スキルを探して」「find skills」「こういうスキル無い?」「スキルで自動化できない?」時に使用。見つからなければ skill-creator で自作する道を示す。ユーザーの確認なしにインストールしない。

ktaroabobon Bundle

ktaroabobon/agent-skills リポジトリにスキルを新規作成・改善する。Claude Code の機能(frontmatter の hooks / allowed-tools / context fork / 動的コンテキスト注入 / 同梱スクリプト)を取りこぼさず組み込み、frontmatter と構造を検証してから公開まで導く。「スキルを作りたい」「skill を新規作成」「このワークフローをスキルにして」「SKILL.md を書いて」「既存スキルを改善したい」「スキルに hook を追加したい」時に使用。スキルの話が少しでも出たら使う。

nxs-cafi Skill

Agentes de IA: LangGraph4J, SpringAI, LangGraph Python, LangChain. Use para criar agentes RAG, grafos de estado, tool calling, multi-agent, human-in-the-loop, ou testar/observar pipelines de LLM.

nxs-cafi Skill

Orquestrador Java para projetos Spring Boot 3.x (Java 21+). Use quando o usuário pedir para criar, revisar ou refatorar código Java/Spring, configurar dependências Maven/Gradle, aplicar DDD, escrever testes JUnit, configurar observabilidade, ou qualquer tarefa de engenharia backend Java. Acionar para "Spring Boot", "JPA", "Micrometer", "Kafka com Spring", "como estruturar pacotes Java".

nxs-cafi Skill

Analisar requisitos e código para escolher, combinar e implementar RBAC, ReBAC e ABAC com segurança. Usar desde a concepção de aplicações com autenticação, múltiplos usuários, roles, organizações, tenants, ownership, compartilhamento, atribuição de recursos, dados sensíveis, APIs, jobs ou Agentes de IA; ao criar ou revisar autorização em Python ou Java; e ao investigar IDOR/BOLA, escalada de privilégio, vazamento entre tenants ou permissões espalhadas. Acionar mesmo sem solicitação explícita de roles quando o domínio indicar diferentes níveis ou escopos de acesso.

shinpr Bundle

Provides Business Model Canvas, Value Proposition Canvas, and TAM/SAM/SOM market analysis frameworks. Use when assessing Viability risk, sizing markets, mapping value propositions, conducting competitive analysis, or identifying market gaps.