AI & ML

igdigitallab Skill

Write song lyrics and a matching style-of-music description for Suno AI (or similar text-to-music tools): structure, meta-tags, vocal styles, ad-libs, rhythm formatting, effects. Use when someone wants a song, lyrics, an anthem, a jingle, a rap, background music for a video, or asks to 'write lyrics / a style prompt for Suno.'

manavdharora Bundle

Spawn 5+ Claude instances into a shared conversation room where they debate, disagree, and converge on solutions. Uses round-robin turns with parallel execution within each round. Triggers on "model chat", "multi-model debate", "agent debate", "spawn a chat room", or /model-chat. Pass a topic as the argument.

manavdharora Skill

Spawn multiple agents into a shared conversation where they debate, disagree, and converge on a solution via a shared chat.json file. Use for architecture decisions, code review debates, design trade-offs, or any problem that benefits from adversarial reasoning. Triggers on "chatroom", "agent debate", "have agents discuss", "agent chatroom", "debate this", or /agent-chatrooms. Also triggers on phrases like "have 3 agents argue about", "multi-agent discussion", or "let them debate".

manavdharora Skill

Define success, constraints, output format, AND explicit failure conditions in a single structured contract. The agent treats it as a spec, not a suggestion. Triggers on "prompt contract", "write a contract", "define success criteria", "failure conditions", "spec this out", or /prompt-contracts. Also triggers on phrases like "what does done look like", "define the contract", "structured spec", or "goal constraints format failure".

manavdharora Skill

Force the agent to ask clarifying questions before starting work. The agent surfaces its own assumptions, the user disambiguates, and only then does it proceed. Triggers on "reverse prompt", "ask me questions first", "what questions do you have", "clarify before starting", "surface assumptions", or /reverse-prompting. Also triggers on phrases like "don't start until you ask me", "what do you need to know", or "ask before building".

manavdharora Skill

Orchestrate parallel browser automation using multiple Chrome DevTools MCP instances. Use when a task requires doing the same browser action across many targets simultaneously (e.g., submitting contact forms, filling applications, scraping pages that need JS rendering). Spin up 1-5 parallel Chrome agents.

manavdharora Skill

After completing a task, spawn reviewer agents to check the output for correctness, edge cases, and simplification. If issues are found, spawn a resolver agent to fix conflicts. Produces 2-3x quality improvement via agent-reviews-agent architecture. Triggers on "verify this", "review loop", "verification loop", "self-review", "agent review agent", "double-check with a subagent", or /subagent-verification-loops. Also triggers on phrases like "have another agent review this", "get a second opinion on this code", or "review and resolve".

manavdharora Skill

Spawn N agents with the same prompt (slight framing variations) to independently analyze a problem, then aggregate results by consensus. Use for decision-making, ranking options, strategic analysis, or any problem where you want to filter hallucinations and surface high-variance ideas. Triggers on "consensus", "poll agents", "stochastic consensus", "spawn N agents to analyze", "multi-agent vote", or /stochastic-multi-agent-consensus. Also triggers on phrases like "what do 10 agents think", "get multiple opinions", or "poll on [topic]".

mine-fnl Skill

Use when reviewing a pull request, a code diff, or a single function for correctness, security, and maintainability. The skill covers Python, TypeScript, Go, and Rust specifically — it cites the language's idiomatic patterns (e.g. error wrapping in Go, Option/Result in Rust) and flags anti-patterns the model would otherwise miss on a generic review. Do not use when the work concerns UI design, documentation copy, or non-code artefacts — pair with a different specialist for those.

mgoodness Skill

Preserve repo-wide agent context docs (AGENTS.md, CLAUDE.md, CONTEXT.md, docs/adr, or similar design/domain docs) on a dedicated branch off the default branch, separate from feature-branch code, so they can be pulled into other worktrees. Use when the user wants to preserve context docs across worktrees, keep design docs out of feature-branch or default-branch commits, or sync a doc correction back onto that branch.

sayonari Skill

Slack MCP サーバの追加・認証（Claude Code 用）

mine-fnl Skill

Use when a customer-support agent is stuck on a conversation, when reviewing a recorded support call transcript for tone, or when coaching a junior agent on how to handle a specific scenario (refund disputes, billing errors, escalation, account recovery). The skill covers the de-escalation ladder (acknowledge → clarify → solve → confirm), empathy without sycophancy, and when to escalate vs. when to push back. Do not use when the issue requires legal review, when the customer is threatening litigation, or for product bug investigation — escalate those immediately.

riyankothari Skill

Run a bounded self-correcting loop that re-feeds the same prompt every time Claude stops, until a completion promise is genuinely true or an iteration cap is reached. Use when a task has a machine-checkable finish line such as a passing test suite or a clean linter, and iterating unattended beats steering each step by hand.

riyankothari Bundle

Cut Claude Code token cost through model-tier routing, budget-capped memory recall, and evidence-gated self-scoring. Use when API cost matters, when deciding whether to delegate work to a cheaper model, when you need memory that survives across sessions without bloating context, or when auditing what is eating the context window.

riyankothari Skill

Keeping credentials out of files, stores and repositories — redacting at write time, scanning before committing, and handling a leak that already happened. Use before committing or publishing anything, before storing prompt or conversation text, when initialising a repo, and whenever a tool persists user input to disk.

rabitarochan Bundle

Structures and generates reusable Skills (SKILL.md) from knowledge discovered during conversations. The core skill of SkDD (Skill Driven Development). Use when: the user says "skillify", "harvest", "/harvest", "turn this into a skill", or "make a skill from this"; when the agent detects a skill candidate at task completion and the user approves; when updating an existing skill with new insights; when promoting a Proto-Skill to a full skill. Also consult for topics related to SkDD, knowledge management, or tacit-to-explicit knowledge conversion.

derek-ye Skill

Adversarial simplify loop — iteratively simplifies changed files with an Editor agent checked by a Reviewer agent. Bash-enforced max 3 epochs.

soukupl Skill

This user's model-and-cost strategy. Use at the start of any non-trivial or multi-step task, when running on Fable, when deciding which model handles a job, and before spawning sub-agents or workflows. Prefers ultracode orchestration on Fable, picks the strongest suitable model per sub-task, and keeps token usage minimal.

soukupl Skill

Keep an up-to-date map of the project's architecture and domain model. Use when adding a new module/service/domain concept, when relationships between parts change, when onboarding into an unfamiliar repo, or when the user asks about "architecture", "the domain model", or "how this fits together". Pairs with the domain-modeling skill.

gabriel2772 Skill

Gera relatórios executivos semanais da Tequaly (KPIs, devoluções, RNCs, desmobilizações, auditorias, inventário) em 30 segundos. Use quando chegar segunda-feira ou fim de semana e você tiver [DADOS] novos para colar. Entrega relatório estruturado em texto ou PPTX com padrão Tequaly, pronto para envio à liderança. Prompt reutilizável toda semana — mesma estrutura, dados novos.

gabriel2772 Skill

Biblioteca de pedidos prontos para as demandas recorrentes de Marcos, para ele não recomeçar do zero a cada vez. Use quando ele mencionar uma demanda que se repete — "relatório semanal de devoluções", "atualiza o dashboard de andaimes", "comunicado de desmobilização", "e-mail pro cliente", "briefing da visita técnica" — ou quando pedir "me dá o prompt pronto disso", "tem um modelo pra isso", "como eu peço isso de novo". Entrega o pedido completo e bem estruturado, já preenchido com o contexto da Tequaly, pronto para executar ou para Marcos colar em outra conversa (Claude ou ChatGPT). NÃO invente dados nos modelos — deixe campos a preencher claramente marcados.

gabriel2772 Skill

Camada profissional sempre ativa para o trabalho de Marcos na Tequaly Técnica Industrial (coordenação de campo, auditoria, logística de andaimes, desmobilização, relatórios para liderança). Use SEMPRE que o pedido tiver caráter de trabalho — relatório, tabela, e-mail, documento, análise, planilha, resumo para diretoria, RNC, auditoria, plano de ação — mesmo que o prompt seja curto, vago ou informal ("faz um relatório disso", "monta uma tabela", "deixa mais profissional", "resume pra diretoria"). Define tom, estilo de saída, proatividade hierarquicamente segura e separação entre fato, hipótese, risco e sugestão. NÃO use em conversa casual ou pergunta de conhecimento geral.

gabriel2772 Skill

Melhora automaticamente prompts profissionais curtos, vagos ou mal estruturados ANTES de executar — e potencializa o resultado. Use SEMPRE que o pedido for de trabalho e o prompt vier simples ou incompleto ("faz um relatório disso", "monta uma tabela", "analisa esses dados", "vê se tem algo errado", "deixa mais profissional", "resume pra diretoria", "transforma em PDF"). Interpreta a intenção real, completa lacunas com suposições prudentes sinalizadas, escolhe o nível certo de profundidade (0 a 3) e executa direto — sem mostrar o processo. Só exibe o prompt reformulado se o usuário pedir ("melhore meu prompt", "me dá um prompt pra colar em outro chat"). NÃO infle tarefas simples nem use em conversa casual.

wolf123450 Bundle

Use when setting up a new Godot 4.x C#/.NET project for Claude Code agentic development, or bootstrapping MCP servers, local docs indexing, third-party skill curation, custom skills, subagents, and hooks for an existing Godot project. Not for API reference lookups — see the docs-layer setup for that.

keetanetwork Skill

Apply explicit advisory limits, recipient controls, and human approval gates around Keeta wallet actions. Use before allowing an agent to quote, convert, bridge, send, or pay out.

kaisa-kucherenko Bundle

Turns a closed ADR into a phase spec an LLM implementer executes phase by phase: cuts, orders and describes — decides nothing (the ADR did), implements nothing (the implementer agents do). Triggers: "/spec", "спека", "напиши спеку", "спека фаз", "нарізка фаз", "write the spec", "phase spec". Requires an ADR with no decision left PENDING (each DECIDED or DROPPED) and its review closed — without one, stop and point at /adr. Running the phases (briefing, gate, review, commit) is /implement, not this skill.

lead21pc Bundle

Audit worldbuilding claims against a project's declared sources, authority model, and internal rules. Use for source tracing, contradiction review, provenance checks, or unresolved lore questions; remain read-only unless the user explicitly requests source changes.

mikhin Bundle

Score the changed files of the current branch with Jev (typesafe-ai evaluation model via Vercel AI Gateway) — bug risk, logic in components, missing tests, published-snapshot writes. Triggers — "/jev", "run jev", "jev the diff", «прогони jev», «что скажет jev».

wolf123450 Bundle

Use when the user wants to set up, install, or replicate a Claude Code status line showing model name, session token usage, context-window percentage, and 5h/7-day rate-limit usage — including on a new machine or in a fresh environment.

yosoyjavieruiz Skill

Use the host-side `agent-browser` CLI for local browser smoke tests, screenshots, snapshots, and simple UI validation against forwarded localhost URLs.

yosoyjavieruiz Skill

Audit or refine ShelraCode's main OpenTUI coding workspace when chat activity, live status, plan/context sidebar, agent visibility, or verification UX must match a supplied visual specification. Do not use for onboarding, settings, or unrelated screens.

comedianhhh Skill

Scaffold the private career/ workspace that every other job-agent skill reads — facts.md (single source of truth), blocks.md, stories.md, rules.md, answers.yaml (standing form answers), targets.yaml (boards to scan), tracker.md, and the per-application template. Use when the user types /career-init, asks to set up job-search files, or when another skill finds no career/ folder.

ldphuong-vn Skill

Business Model (Stage 5)

matthewtaormina Bundle

Official guide for developing with multi-model AI gateways and routers, specifically OpenRouter (global auto-routing and multi-provider failover) and EuroRouter / EUrouter (sovereign EU data residency and GDPR compliance). Use when implementing multi-model fallbacks, dynamic routing, or EU-sovereign LLM infrastructure using OpenAI-compatible SDKs.

matthewtaormina Bundle

Official guide for developing with the Anthropic SDK (@anthropic-ai/sdk on npm and anthropic on PyPI) and Claude Messages API. Use when implementing, upgrading, or debugging Anthropic API calls, Claude 3.7 Sonnet, Claude 3.5 Sonnet/Haiku, prompt caching, extended thinking, streaming, tool use/function calling, token counting, message batches, or multimodal processing.

ruslankhis Bundle

Help build, change or review a Python Google ADK agent by selecting the relevant Agentic Engineering specialist skills. Use as the starting point for an ADK engineering request when the user has not chosen a specialist, or when a change spans several ADK concerns. For a focused task already covered by an active specialist, continue with that skill directly.