3. Data, Science & AI

Updates and maintains memory bank files including active context, progress, and decision logs for AI agent memory management.

Instantly computes mathematical operations such as derivatives, integrals, limits, series, and probability without requiring conversation.

Generates and organizes YouTube transcripts into NotebookLM-compatible source files with automatic merging, CSV support, and retry functionality.

Searches scientific formulas, constants, and models from external databases including Wikidata and BioModels.

Enables AI-driven image editing in ComfyUI using Qwen 2.5VL for image analysis and camera parameter adjustments.

Manages conversation session memory by loading context, summarizing sessions, compressing old sessions, and enabling memory search for efficient AI interactions.

Provides best practices for data engineering, including ETL/ELT patterns and reproducible data workflows for pipelines and infrastructure.

Provides analytics frameworks for analyzing performance metrics, creating reports, and evaluating growth experiments using KPIs and benchmarks.

Enables design and implementation of agent-native applications where autonomous agents are central, handling tasks through iterative loops and self-modification.

Enables low-level Gemini API control for generating images with custom resolution, aspect ratios, and multi-turn refinement for non-brand visual assets.

Validates database migration PRs for data integrity, ensuring ID mappings, column renames, and data transformations are safe and correct against production data.

Generates AI-optimized context for NuGet packages to enhance documentation, API references, and dependency analysis.

Generates pre/post-deployment checklists with SQL verification queries, rollback procedures, and monitoring plans for high-risk data changes in production.

Reviews database migrations and data model changes for integrity, constraints, and referential safety.

A Ruby framework for building type-safe, composable LLM applications with signature creation, model configuration, and agent systems.

Crafts optimized prompts for LLMs including system prompts, user prompts, and few-shot examples to enhance model performance.

Creates optimized meta-prompts for Claude-to-Claude pipelines, enabling research, planning, and execution stages to produce consumable outputs for subsequent AI processing.

Guide to adversarial machine learning techniques including adversarial examples and data poisoning.

Manages the full lifecycle of AI agent skills including creation, optimization, versioning, and deprecation for efficient AI capability deployment.

Data science and machine learning platform for East language, enabling optimization, model training, and explainability with libraries like XGBoost and SHAP.

Manages Google Tag Manager configurations for analytics, conversion tracking, and event management through container, tag, trigger, and variable setup.

Processes INQ QUESTION.md files to generate research agent prompts for Phase 1 deliberation.

Enables safe querying of production and staging Supabase databases with approval workflows, safety checks, and audit logging for data verification and debugging.

Provides seamless integration with Google Gemini API and orchestrates AI model workflows for efficient deployment and management.