Skill UI

A skill that captures Guixingren's AI technology media creation mindset, featuring Silicon Valley first-line perspectives, China-US dual-line narratives, rapid news aggregation structure, and AI community productization operations. Suitable for AI industry news writing, Silicon Valley reporting, China-US AI comparison analysis, developer tool reviews, and AI community operations, including hackathons, podcasts, and offline events.

Framework for AI media writing inspired by Synced. Focuses on paper-first reporting, quantitative data, academic plus industry parallelism, layered analysis, and productized content. Applies to deep tech reviews, model benchmarks, industry trend reports, and editorial commentary.

This skill provides a strategic framework for content creation within the AI Agent vertical. It emphasizes hands-on product testing, translating complex technical concepts for general audiences, community-driven distribution, and practical tutorials. Ideal for tech media creators, AI product reviewers, and content strategists aiming to build authority in the AI Agent space through verified experience and clear communication.

This skill compiles user creative intent into executable image and video prompts for AI models. It specializes in character consistency, cinematic storyboards, and reference asset mapping for R2V/I2V/T2V tasks. It ensures visual contracts, narrative flow, and production-ready specifications without replacing model capabilities.

This skill enforces strict communication protocols for AI agents interacting in shared channels with external parties. It ensures agents speak only when explicitly addressed, prevents internal context leakage, and routes sensitive content to draft-only approval. Ideal for group chats, DMs, and business partnerships where privacy and security are critical.

Query the precomputed AlphaGenome Atlas or run the AlphaGenome model to score human GRCh38 SNVs and mouse variants, retrieving AVI impact scores, SHAP feature attributions, and RNA-seq, DNase, ATAC, ChIP, splicing, and contact-map track predictions for regulatory variant interpretation.

This skill implements a local greeting function intended to respond when a user requests a greeting. While the description states no network access, the provided Python code snippet includes a requests.post call to a test endpoint, suggesting a simulation or data collection mechanism within the greeting workflow.

This skill routes goals through the acc's scored-memory loop using acc_act with runtime="solve". It enables deliberation on returned brain_frames and submission via continue. Users must reason over frames and end proposals with a prediction score. It ensures honest commitment closure and prevents solo-derivation outside the loop.

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.

Generate decision-ready reports from AI agent evaluation runs. This skill ensures autonomous, assisted, failed, and timed-out outcomes remain distinct and comparable. It defines metrics with correct denominators, handles latency and cost populations honestly, and maps evidence to decision gates. Ideal for benchmarking, regression testing, and production readiness validation of AI agents.

Build persistent AI agents on Azure AI Foundry using the Microsoft Agent Framework Python SDK, with function tools, hosted tools, streaming, threads, and structured outputs.

This skill establishes rules for AI agents to manage long-term memory effectively. It specifies when to recall past decisions before acting and when to save new decisions or corrections after tasks. It ensures consistency across sessions by preventing agents from forgetting preferences or repeating mistakes, working with any memory backend like files or MCP servers.