3. Data, Science & AI

Provides best practices for crafting effective prompts for AI agents and LLMs, specifically within the OpenPromo codebase using AI SDKs.

Converts session logs from markdown format to structured JSON schema for system consistency and automated processing.

Enables deep traversal of knowledge graphs to uncover relationships between entities and provide comprehensive context for planning and analysis.

Optimizes system prompts for Claude Code agents using proven patterns to enhance agent performance and behavior.

Provides a unified four-tier memory system for AI agents (semantic, episodic, causal) enabling memory-first architecture per ADR-007.

Collects and reports agent usage metrics from git history, implementing 8 key metrics for system health, effectiveness, and adoption as defined in agent-metrics.md.

Guides curation of knowledge memories by updating outdated content, marking obsolete items, and building knowledge graph connections for enhanced information quality.

AI-powered tool that analyzes input to recommend, improve, or create new skills using multi-agent synthesis and deep iterative analysis.

Manages database operations including schema migrations, SCD Type 2 historical tracking, and Closure Table hierarchical data structures.

Provides advanced data modeling patterns including SCD Type 2 for historical tracking, Closure Table for hierarchical data, and budget management structures.

Guides management of the Agent Registry, a database-backed AI agent configuration system via UI, API, CLI, and MCP tools.

Provides tools for creating and executing AI agent behavior evaluations, including YAML case definitions, assertions, and LLM-as-judge scoring.

Generates consensus outputs via recursive multi-LLM model battles, refining results through iterative competitive evaluation.

Enables iterative AI content refinement through multi-model competition and critique, rotating between OpenAI-compatible and Ollama models.

Fetches OpenRouter model details and provides configuration guidance for integrating models into acai-ts provider setup.

Provides efficient patterns for launching and managing parallel AI subagents, optimizing resource utilization and task coordination.

Generates full-stack AI applications using Next.js, AI SDK, and ai-elements for chatbots, agent dashboards, and custom AI solutions.

Enables building AI agents, chatbots, and streaming applications with Vercel AI SDK v6 for seamless AI integration.

Enables semantic, vector, and hybrid search in PostgreSQL using pgvector and ParadeDB, supporting indexing, re-ranking, and RAG implementations.

Enables development of AI agents, multi-agent workflows, and tool integrations using the OpenAI Agents SDK in Python for streaming applications.

Provides persistent cross-session knowledge management, serving as the single source of truth for project context in AI applications.

Retrieves historical knowledge from Project CLAUDE.md, User CLAUDE.md, and Memory MCP for AI-assisted applications.

Manages autonomous agent lifecycle including spawning, cancellation, listing, and result checking via specific commands.

Creates specialized AI agents for codebase areas with knowledge bases, sub-commands, and codebase integration capabilities.