Explorar skills

Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.

Set up visual regression testing with baseline screenshots across breakpoints. Auto-detects frontend framework (Next.js, React, Vue, Angular, Flutter, Storybook), configures Playwright screenshot comparison, BackstopJS, or Flutter golden tests, captures every page at mobile/tablet/desktop/wide viewports plus interactive states (hover, focus, error, empty, loading), stabilizes dynamic content to pr

Comprehensive guidance for planning, executing, and optimizing B2B paid media campaigns across search, social, display, CTV, OOH, and ABM channels — trigger when a user needs help with paid advertising strategy, channel selection, budget allocation, creative approach, audience targeting, or campaign measurement.

Inspect and analyze codebases using pygount for LOC counting, language breakdown, and code-vs-comment ratios. Use when asked to check lines of code, repo size, language composition, or codebase stats.

Set up a complete mobile publishing pipeline — chains CI/CD workflow generation, iOS App Store publishing via Fastlane, Google Play Store publishing, and analytics verification into an end-to-end release automation. Use for Flutter, React Native, or native iOS/Android apps ready for store submission.

Activates DatabaseGenius for database design, optimization, and migration. Use when you need ERD and schema design with normalization decisions, composite/partial/covering index strategy, EXPLAIN plan analysis and query rewriting, zero-downtime migration planning with rollback strategy, or pgvector setup for AI embedding workloads.

Use this skill when promoting Jiang Lens proposals, glossary entries, model claims, ledger items, or chronology-sensitive material into canonical content after source grounding and review records exist.

Generate images, videos, audio, and more using fal.ai AI models. Use when user requests: "generate image", "create video", "make a picture", "text to image", "image to video", "text to speech", "transcribe audio", "edit image", "remove background", "upscale image", "enhance resolution", "search models", "fal pricing", "fal usage", "create workflow", "chain models", "setup fal", "add API key", or s

Activates VoiceAgent — a specialized engineer for designing and building voice AI agents. Use when you need to build speech-to-text pipelines, text-to-speech integrations, real-time voice assistants, telephony bots (Twilio/Vapi), or wake-word detection systems with low-latency streaming and interruption handling.

Dynamic pricing engine for short-term rentals (Airbnb, VRBO, Booking.com, Direct) — sets nightly rates based on a comp set (40+ filters: bedrooms, bathrooms, amenities, location radius, rating tier), seasonality patterns (weekly + monthly + annual cycles per market), local events (concerts, conferences, sports), lead-time demand curves (60+ day, 30-day, last-minute), booking pace (year-over-year s

Comprehensive K-12 or higher-ed system analysis: predict student dropout risk with early warning indicators, optimize curriculum alignment and pacing, personalize learning paths for at-risk and diverse populations, and audit school operations for resource efficiency. Use when building or auditing an SIS, LMS, student success platform, or education data system.

Envie as alterações de código da sessão atual para a CLI do OpenAI Codex para revisão. O Codex revisa as diferenças em busca de bugs, segurança, desempenho e qualidade.