Explorar skills

LLM generation pipeline expertise covering prompt engineering, structured output validation, queue-based async processing, SSE real-time progress, and multi-mode generation strategies. Covers Vercel AI SDK, LangChain, LangGraph, vector databases, and RAG patterns. Trigger keywords: AI generation, LLM pipeline, prompt engineering, structured output, Zod schema, BullMQ, SSE, RAG, vector database, em

Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking. Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks.

Traga sua própria chave de API OpenAI, Google ou OpenRouter para o DeepRead. Todo o processamento de documentos é roteado através da sua conta, sem custo de LLM do DeepRead e ignorando totalmente a cota de páginas, usando a mesma API e endpoints.

Atualiza o arquivo llms.txt para refletir mudanças na documentação. Use ao editar detalhes ou especificações do repositório; para criar do zero, consulte create-llms.

Classificação e agrupamento de UP's do Bilibili seguidos, com sincronização assistida por IA.

Otimização de tokens pronta para produção reduz custos em 40-75% com poda de recuperação, cache inteligente e roteamento de modelos. Ideal para otimizar custos de API, latência ou gerenciar contextos longos, especialmente em pipelines RAG, sistemas de alto volume, conversas multi-turno ou quando o contexto excede 2K tokens.

Design LLM applications using the LangChain framework with agents, memory, and tool integration patterns. Use when building LangChain applications, implementing AI agents, or creating complex LLM workflows.

Use when setting latency budgets per request type, when tracking LLM cost per call type, when designing a caching strategy, when reviewing bundle size, when profiling a slow path, or when a service is approaching its latency or cost SLO.

Full project onboarding: generates a lean CLAUDE.md, bootstraps LLM Wiki, links Obsidian vault, and installs eagle-skills (11 skills + 12 agents). Use when starting a new project, onboarding to a codebase, or when the user says: 'init claude', 'setup claude md', 'create claude.md', 'project init', 'eagle init', 'bootstrap this project', 'set up this repo for claude', 'add claude.md', 'onboard this

Batch translate JSON locale files to 39+ languages using LM Studio (local LLM). Structure-preserving translation with validation. Requires i18n-translator extension. Triggers on: /creo i18n, translate, localize, i18n, internationalization.

Audit an existing .bib file entry-by-entry against ground truth (DOI, Semantic Scholar, OpenAlex, arXiv) to catch the silent citation errors that pass spell-check but kill peer review — wrong year, mis-cited authors, wrong journal/volume, swapped title-author pairs, fabricated DOIs, and outright hallucinated entries (a real concern with LLM-assisted drafting). Spawns one narrow-focus subagent per

This skill should be used when the user wants to "humanize this", "make this sound less AI", "rewrite this naturally", "remove AI patterns", "make this more conversational", "this sounds robotic", "edit for voice or tone", or pastes text that reads like LLM output (parallel structure in threes, em dashes, semicolons, hedging phrases, hollow affirmations). Also trigger for "make this sound like me"