3. Data, Science & AI

Enables integration of OpenAI's GPT-5.2 and GPT-5-mini models via AI SDK 6 for complex and fast AI tasks.

Automatically enhances capabilities by learning from interactions and avoiding recurring anti-patterns.

Expert debugging guide for AI Counsel's convergence detection system, resolving non-converging deliberations, early stopping failures, and semantic conflicts.

Analyzes AI Counsel decision graphs to uncover historical deliberations, detect patterns, and resolve memory-related issues.

Retrieves contextually relevant information from a LlamaCloud Index using Retrieval Augmented Generation (RAG) with llama_cloud_services.

Manages LlamaIndex workflows as LlamaAgents via CLI, enabling initialization, local preview, deployment, and workflow management.

Enables accurate document text extraction using LlamaParse, requiring llama_cloud_services package and API key configuration.

Processes data files through Python and shell scripts for analysis and transformation.

Provides a unified API for seamless file operations across local, S3, HDFS, and cloud storage systems, simplifying data access in data workflows.

Provides a procedural memory system for AI coding agents using three-layer cognitive architecture with confidence decay and safety features.

Enables inter-agent messaging within AI Maestro's system for collaborative task execution between AI agents.

Generates images from text or reference using OpenAI and Google AI APIs with customizable aspect ratios and quality settings.

Coordinates subagent task distribution, collaboration, and autonomous execution mode with dynamic scale determination for AI systems.

Coordinates sub-agent task distribution and autonomous execution, managing scale judgment for efficient large-scale task division.

Performs quality control on single-cell RNA-seq data using scverse best practices, including MAD-based filtering and visualizations for data assessment.

Executes nf-core bioinformatics pipelines for RNA-seq, WGS/WES, and ATAC-seq data analysis using local FASTQs or public datasets (GEO/SRA).

Provides deep learning tools for single-cell omics data analysis, including batch correction, multi-modal integration, and specialized methods (scVI, totalVI).

Converts laboratory instrument files (PDF, CSV, Excel, TXT) to Allotrope Simple Model (ASM) JSON or flattened CSV for LIMS, data lakes, and downstream analysis.

Designs efficient database schemas including entity relationships, normalization, and structure selection for relational or NoSQL databases.

Processes and transforms CSV data using command-line tools for efficient data manipulation and analysis.

Optimizes AI prompts using 57 proven frameworks to enhance model responses and task effectiveness.

Executes recursive language model loops for extended context handling, leveraging a local Python REPL and subagent for sub-LLM queries.

Defines memory retrieval and storage protocol for AI agents, ensuring memory is foundational to session reasoning from initialization.

Offloads context-intensive tasks such as codebase exploration and documentation generation using Codex subagents in background processes.