Skill UI

TensorBoard delivers a unified interface to track training metrics, debug network behavior, compare experiments, inspect model graphs, and profile performance for PyTorch or TensorFlow workflows.

Assists with tensorboard visualizer tasks for ML training by guiding setup, best practices, and experiment tracking whenever tensorboard visualization is mentioned, producing production-ready configs and validated outputs.

Weights & Biases (W&B) automates experiment logging, real-time dashboards, hyperparameter sweeps, artifact tracking, and model registry/versioning for ML teams using PyTorch, TensorFlow, or HuggingFace workflows.

Hierarchical Agent Memory (HAM) scopes Claude Code agent context by generating per-directory CLAUDE.md files, tracking token savings, and surfacing analytics via a dashboard so multi-folder projects avoid repeated full scans.

A comprehensive protocol defining communication rules and orchestration patterns for multi-agent C-suite teams. It structures complex cross-functional analyses, simulates board meetings, and enforces hard rules—such as depth limits and loop prevention—to ensure consistent, verifiable, and non-circular business insights.

Simultaneously activates multiple expert agents to analyze any topic through parallel perspectives, then synthesizes their insights into one decisive recommendation for complex product, strategy or policy questions.

Trackio is a robust library designed for comprehensive monitoring and management of Machine Learning experiments. It allows users to log key training metrics, set structured alerts (e.g., loss divergence, NaN gradients), and visualize results over time. The system supports integration via Python API for logging/alerting, and a powerful CLI for programmatic metric retrieval and dashboard synchronization to Hugging Face Spaces. Ideal for autonomous AI agent workflows and detailed research tracking.

Turns your existing Kanban board into an autonomous AI-driven development pipeline, coordinating Claude Code workers, deterministic gates, cost tracking, and crash-proof orchestration for full lifecycle work.

Guides a warm adaptive onboarding conversation to build a personalized SOUL.md for an AI partner, capturing the user’s identity, role, preferences, and vision across four phases before generating and confirming the file.

Interactive workflow that interrogates scope, deliverables, constraints, and success criteria while guiding users to refine task prompts; copies the final markdown with Joyride clipboard commands and never writes code.

Automate Datadog monitoring tasks via Rube MCP’s Composio toolkit—query metrics, search logs, manage monitors/dashboards, and create events or downtimes while respecting live tool schemas.

Napkin gives Copilot CLI users a browser whiteboard to draw, sketch, and add sticky notes; Copilot reads snapshots and JSON data, then replies with analysis, suggestions, and next steps for business-oriented thinkers.