Skill UI

Use the Recommendation Canvas to vet AI product concepts by mapping business and customer outcomes, hypotheses, positioning, risks, and success metrics so you can make defensible investment recommendations.

A comprehensive guide to building production-grade evaluation systems for Large Language Models (LLMs). This skill covers advanced methodologies, including Direct Scoring and Pairwise Comparison, alongside critical techniques for mitigating systematic biases (e.g., Position Bias, Length Bias). Learn how to select appropriate metrics and structure prompts to ensure reliable, consistent, and objective quality assessment across various AI applications.

This comprehensive guide outlines best practices for building reliable and robust AI agent harnesses. It covers optimizing action spaces, defining structured tools, managing observation formats, and implementing essential error recovery contracts. Use these principles to dramatically improve agent completion rates, manage context budgets effectively, and transition AI prototypes into stable, production-grade autonomous systems.

This protocol defines the structured communication rules for multiple specialized AI agents simulating C-suite roles (e.g., CFO, CTO, CRO). It dictates invocation syntax, response formatting, and enforces strict constraints (such as loop prevention, call depth limits, and isolation rules) to ensure coordinated, non-contradictory, and reliable cross-functional analysis for executive decision-making.

This pattern defines a robust, continuous loop structure for autonomous agents. It integrates crucial components like quality gates, evaluation harnesses, and explicit recovery controls. It guides complex workflows by defining decision paths (e.g., PR control, RFC decomposition) and includes strategies to mitigate common failure modes such as infinite loop churn or cost drift, ensuring measurable progress in autonomous systems.

This pattern addresses the fundamental limitation of context window size by treating the filesystem as a persistent, external memory. Agents can write large outputs, plans, or findings to files, offloading them from the active context. Instead of carrying all information, agents use targeted search and retrieval tools to load only the necessary context on demand, drastically reducing token bloat and improving performance over long, complex tasks.

This infrastructure provides a robust solution for running powerful AI agents in remote, sandboxed environments. It overcomes the limitations of local execution by ensuring unlimited concurrency, consistent environments, and collaborative features. Use this to build complex background coding agents, multi-client interfaces (e.g., Slack, Web), or scale agent workloads far beyond local machine constraints.

This skill provides comprehensive tools for documenting and managing model performance on Hugging Face. It supports multiple methods for adding structured evaluation data, including extracting tables from README content, importing benchmark scores via APIs (like Artificial Analysis), and executing high-performance custom evaluations using advanced backends such as vLLM or lighteval. It ensures that all metadata is structured and compatible with model leaderboards and industry standards.

This skill provides comprehensive tools for AI researchers and engineers to publish, index, and manage academic papers directly on the Hugging Face Hub. It supports automatic arXiv indexing, linking papers to specific models and datasets, verifying authorship, and generating professional, markdown-based research articles. It streamlines the entire scientific publication workflow from creation to professional publication.

Analyze mental health data to track mood, anxiety, depression scores, therapy progress, and crisis risk. Supports correlation analysis with sleep, exercise, nutrition, and chronic conditions to generate personalized insights and recommendations.

Guides building SEO-optimized pages at scale by combining templates, structured data, and proprietary insights, helping teams choose the right playbooks, validate demand, and keep each page unique and valuable while avoiding thin-content penalties.

This skill formalizes successful, recurring patterns discovered in auto-memory (MEMORY.md) and graduates them into permanent, enforced rules within the project's core instruction set (CLAUDE.md or scoped rules). It ensures that learned best practices become mandatory project conventions, significantly improving adherence and consistency across all future AI interactions and development cycles. Use this when a temporary note proves to be a vital, repeatable guideline.