3. Data, Science & AI

Evaluates codebase maturity across eight technical pillars to assess readiness for autonomous AI development and identify gaps.

Guide for implementing database solutions using Prisma, SQLite, and LiteFS in the Epic Stack framework.

Runs Tecton feature platform tests and plans via Pants in data-science repositories with full output capture for long-running commands.

Enables building conversational AI, chatbots, and text generation features using LLM chat completions with context management.

Generates videos from text prompts or images using an AI SDK, supporting asynchronous task management for programmatic video content creation.

Generates images from text descriptions using an AI SDK, supporting multiple sizes and base64 output. Includes CLI for quick image creation.

Enables image analysis and visual content description using vision-language models via z-ai-web-dev-sdk, supporting URLs and base64 images for multimodal AI chat applications.

Transcribes audio files to text using automatic speech recognition (ASR) via z-ai-web-dev-sdk, supporting base64 encoded inputs.

Converts text to natural-sounding speech using AI-powered TTS with voice and speed customization.

Extracts structured web content including titles, HTML, and metadata using AI-driven scraping capabilities for data processing.

Retries failed AI queries using a higher-capacity model to improve response accuracy and reliability.

Generates an HTML dashboard to visualize routing statistics, providing actionable insights into application or network traffic patterns.

Routes user queries to the most suitable Claude model (Haiku, Sonnet, or Opus) based on query complexity and performance requirements.

Provides agentic orchestration patterns for long-running tasks, implementing evidence-based delivery and Simon Willison's agent loop for multi-step workflows and agent coordination.

Tool for designing, optimizing, and quality-checking AI prompts based on Anthropic's guidelines, supporting creation, refinement, and best practices.

Assists in developing, training, evaluating, and deploying machine learning models with MLOps practices.

Provides stuck detection, memory management, and session learning for AI agents to enhance operational efficiency.

Provides a foundational identity framework for AI companions, defining voice, behavior, relational structure, and recovery protocols to maintain consistent interaction.

Executes AI code tasks via multiple backends (Codex, Claude, Gemini) with file references and structured output.

Orchestrates multi-agent AI workflows via Kiro specifications, managing state through JSON and markdown files with command-triggered execution.

Provides structural validation and damage propagation systems for Three.js-based building games with arcade, heuristic, and realistic physics modes.

Simulates particle systems with realistic physics including forces, collisions, and field-based motion for artistic or scientific applications.

Searches indexed text corpora (sessions, docs, logs) with enhanced capabilities beyond traditional grep tools.

Enables multimodal AI interactions with Google Gemini API using @google/genai SDK, supporting text, image processing, and function calling while resolving SDK errors.