Skill UI

OpenViking Memory enables AI agents to recall and persist long-term semantic knowledge using MCP tools. It supports retrieving prior context at task start via find/search/read and storing durable facts, preferences, and lessons via remember. Ideal for coding, debugging, and multi-step workflows, ensuring continuity across sessions without automatic lifecycle hooks.

Persistent file-based planning for multi-step AI agent tasks. Maintains task_plan.md, findings.md, and progress.md on disk, with lifecycle hooks injecting selected plan context. Best for research or work needing 5+ tool calls.

This skill guides users on setting up and running LLM-as-judge evaluations using Arize. It covers defining evaluators (including prompts, classification choices, and models) and executing tasks against real data spans or experiments. It supports key metrics like hallucination, faithfulness, and correctness, and facilitates continuous monitoring for comprehensive LLM performance analysis.

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.

Reads `.agent.md` or `.prompt.md` files, assesses task complexity, tooling, and requirements, then recommends the most cost-efficient GitHub Copilot model(s) (e.g., GPT-5, Claude Sonnet) with rationale for users across different subscription tiers and priorities.

A skill for continuous file-based planning in multi-step AI agent workflows. Maintains task_plan.md, findings.md, and progress.md on disk, with lifecycle hooks injecting project-specific planning context. Supports explicit session catch-up via session-catchup.py for metadata or limited replay, and optional controlled mode. Designed for research or tasks requiring 5+ tool calls.

This skill is designed to manage complex, multi-step tasks by coordinating the efforts of multiple specialized AI agents. It provides structured methods for listing available agents, initiating real-time conversations, submitting complex background tasks (such as data analysis or report generation), tracking task status, and maintaining conversational context using session IDs. It is essential when a core task requires diverse expert knowledge or sequential collaboration.

Dispatch dedicated agents with isolated contexts to handle multiple independent failures or tasks in parallel. Use when you have several unrelated issues that can be investigated separately without shared state, while preserving your own context for coordination and review.

Evolver is a comprehensive self-evolution engine designed for autonomous AI agents. It analyzes complex runtime history, identifies performance bottlenecks, and autonomously generates and applies improvements. Communication is architecturally confined to a local Proxy Mailbox, ensuring secure and reliable interaction with the EvoMap Hub for asset management and task acquisition. Ideal for developing highly adaptive, continuously improving intelligent systems.

Generate and edit images using OpenRouter's AI models like Gemini and Seedream. Supports photos, illustrations, logos, and concept art via command line. Requires API key for generation tasks.

Build clinical and healthcare deep learning pipelines with PyHealth: load EHR, signal, and imaging datasets, define prediction tasks, instantiate models, train, and evaluate using a modular five-stage workflow.