Skill UI

This guide provides advanced techniques to optimize Deepgram API performance. Learn how to preprocess audio using ffmpeg (resampling, noise reduction), select optimal transcription models (speed vs. accuracy), and implement efficient streaming or parallel batch processing to achieve high throughput and low latency.

A comprehensive checklist and guide for deploying Deepgram integrations into a production environment. It covers critical areas including robust authentication practices, system resilience design (e.g., backoff strategies, circuit breakers), performance optimization (singleton clients, audio preprocessing), security measures (PII redaction), and mandatory monitoring setup (Prometheus metrics and alert rules). Essential for any go-live phase.

This skill teaches the critical principles and decision-making processes for safe production releases. It covers platform selection (e.g., Kubernetes, Vercel), mandatory pre-deployment verification (Code Quality, Environment), structured workflows (Prepare, Backup, Deploy, Verify), and advanced strategies like Blue-Green and Canary deployments. Focuses on thinking through risk mitigation, not just executing scripts.

A comprehensive diagnostic tool designed for deep debugging of complex Exa system issues. It systematically checks multiple layers (DNS, Authentication, Neural Search, Content Retrieval, etc.) and provides detailed latency profiling (p50, p95) to pinpoint performance bottlenecks or intermittent failures. Use this when standard troubleshooting steps fail or when preparing detailed evidence for support escalation.

A comprehensive checklist guiding the full lifecycle deployment of Exa search integrations to a production environment. It covers critical stages including pre-deployment security audits, code quality verification, performance baselines, monitoring setup, and emergency rollback procedures. Use this guide when preparing for a major launch or verifying production readiness.

Automatically activates when you mention feature engineering helper, offering guidance on data preparation, model training, hyperparameter tuning, and experiment tracking for ML projects.

Automates guidance on assessing feature importance for machine learning training, covering data prep, model training, tuning, and experiment tracking while generating production-ready code, configurations, and validation instructions for best-practice compliance.

Guides ML engineers through dataset prep, PEFT strategy selection, LoRA/QLoRA adapter setup, training monitoring, and deployment checks for practical LLM fine-tuning, quantization, and adapter merge workflows.

This skill orchestrates a comprehensive, multi-stage Git development workflow, simulating the process from code commit to PR preparation. It leverages specialized AI agents (code reviewers, test automators, prompt engineers) to systematically enforce best practices. Key functions include code quality assessment, breaking change analysis, running unit/integration/E2E tests, and generating Conventional Commit messages, ensuring code readiness and high quality before merging.

This expert tool orchestrates the complete onboarding experience for new team members in technical or remote-first organizations. It provides a structured roadmap covering pre-arrival preparation (access setup, documentation), Day 1 orientation, technical stack immersion (CI/CD, codebases), and cultural integration (buddy system). Use this when you need a comprehensive, step-by-step guide to rapidly transition new hires into productive, integrated team members.

This utility automatically collects comprehensive diagnostic data for the Granola application, including system specifications, audio configurations, network connectivity tests, and application metadata. It packages all findings into a single ZIP file for support teams. Use this when preparing a detailed support ticket or troubleshooting complex, persistent issues. The process is designed to collect diagnostic context without exposing sensitive meeting content, transcripts, or private API keys.

Provides expert guidance for GRPO training with TRL, covering dataset prep, reward engineering, structured outputs, and multi-objective alignment when preference pairs are scarce.