3. Data, Science & AI

Provides a curated repository for discovering, retrieving, and improving AI prompts, enabling effective prompt templates for LLM interactions.

Orchestrates multi-agent AI workflows for content creation, activated by `prose` commands, .prose files, or mentions.

Transcribes audio files into text using OpenAI's Whisper API, enabling accurate speech-to-text conversion for applications.

Provides a command-line interface for Gemini AI model interactions, enabling one-shot Q&A, text summarization, and content generation.

Enables local speech-to-text conversion using OpenAI's Whisper model without API keys, for privacy-focused transcription.

Enables offline text-to-speech synthesis using the sherpa-onnx library, eliminating cloud dependencies for local speech generation.

Summarizes per-model cost usage for Codex and Claude AI models from CodexBar CLI data, providing current or full breakdowns.

Creates and packages AgentSkills with scripts, references, and assets for LLM-based agent workflows.

Provides best practices for oracle CLI usage including prompt management, file bundling, engine selection, and session handling in AI model workflows.

Generates spectrograms and feature-panel visualizations from audio files using the songsee command-line tool for data analysis.

Batch-generates images via OpenAI API with random prompts and creates an HTML gallery for display.

Enables creation, editing, and analysis of spreadsheets including formulas, formatting, and data visualization for data processing tasks.

Adds product analytics events to track user interactions within Metabase frontend for data-driven insights and behavior analysis.

Expert guidance on ClickHouse query optimization, data engineering patterns, and high-performance analytics implementation for analytical workloads.

AI system observing sessions via hooks to create and score atomic instincts, evolving them into skills, commands, and agents.

Pattern for progressively refining context retrieval in LLM-based systems to solve subagent context challenges.

Calculates key financial ratios and metrics from financial statements to support investment analysis and decision-making.

Guidelines for creating and updating modes in PostHog AI agent to control tool usage, prompts, and prompt injections for optimized performance.

Optimizes vector index performance for latency, recall, and memory by tuning HNSW parameters and quantization strategies in AI search systems.

Enables analytics engineering with dbt for organizing data models, testing, documentation, and incremental data transformations.

Builds Retrieval-Augmented Generation (RAG) systems for LLMs using vector databases and semantic search to enable knowledge-grounded AI applications.

Designs PostgreSQL schemas with best practices, including data types, indexing, constraints, and performance patterns.

Designs LLM applications with agent architectures, memory management, and tool integration using LangChain 1.x and LangGraph.

Automates end-to-end machine learning pipeline creation, covering data preparation, model training, validation, and production deployment.