3. Data, Science & AI

Initializes a standardized .agent/ folder structure for AI agent development, including documentation, session tracking, and task management based on production patterns.

Provides deep integration guidance for Polymarket's CLOB and Gamma APIs, enabling market data fetching and order execution for trading applications.

Enforces critical rules for safe database schema changes, migrations, and prevents unauthorized manual database modifications across all agents.

Configures data catalog connectors including Unity Catalog, Databricks, Iceberg, and Glue for seamless data catalog setup and management.

Enables development and deployment of AI features including LLM integration, RAG systems, and machine learning models using frameworks like LangChain and vector databases.

Searches and filters MLflow experiment traces via CLI or Python API for debugging and analysis of machine learning workflows.

Specializes in optimizing data formats including TOON, JSON, and YAML with serialization patterns and validation for modern application data pipelines.

Optimizes LLM prompts and agent instructions using best practices in prompt engineering for improved model performance across all AI platforms.

Guides developers in building AI applications using Vercel AI SDK v6 for LLM features like agents, tool calling, and structured outputs in TypeScript/JavaScript.

Enables end-to-end integration with Synth AI API for model evaluation and GEPA tasks via SDK and HTTP interfaces.

Provides optimized Postgres schema design and query best practices for performance, including Prisma integration guidance.

Guides creation of Model Context Protocol (MCP) servers for LLMs to interact with external APIs, supporting Python (FastMCP) and Node/TypeScript (MCP SDK).

Enables advanced evaluation of LLM outputs through direct scoring, pairwise comparison, and bias mitigation to establish quality standards and compare model performance.

Configures and debugs LLM-powered blog generation using Vercel AI SDK and Google Gemini, including prompt adjustments and integration fixes.

Transforms vague requests into production-ready, hallucination-free prompts for Claude 4.x using anti-hallucination guards and best practices.

Manages currency conversions using the Frankfurter API for USD, EUR, and ILS, enabling multi-currency dashboard displays.

Provides Prisma schema patterns for constraints, decimal precision, soft deletes, and database conventions to streamline schema management and data integrity.

Transforms basic topics into professional LLM prompts using the C.R.A.F.T. framework for optimal model performance.

Converts Nextflow process definitions into Galaxy tool XML format for seamless bioinformatics workflow deployment.

Routes Nextflow workflow definitions to Galaxy conversions, enabling bioinformatics workflow interoperability.

Systematically resolves gaps in LLM applications through iterative user testing, fix implementation, and validation cycles to ensure model performance.

Converts Nextflow pipelines into Galaxy workflows for reproducible bioinformatics data analysis and scientific research.

Expert assistance for PostgreSQL database design, optimization, migrations, and JSONB pattern implementation on the PayU Digital Banking Platform.

Automates creation and updating of LLM context files (CLAUDE.md, .cursorrules, .goosehints) across repositories to enable AI agent codebase comprehension.