Machine learning

Use this skill for internal TypeScript work in the nuxt-ui-tools repository.

Threadlight evals leg: offline batch quality, Foundry Continuous Evaluation and champion-challenger gates.

写中文的时候直接绕开 AI 腔，而不是写完再洗。给 LLM 当写作指令用：喂进题目和文体， 产出的初稿就已经没有模板句式、没有价值拔高骨架、没有等厚段落。 这是六段流水线的第一段。第一段决定文章有没有人味，后面几段只能修，补不出来。 这份文件本身刻意写成它要求的那种中文 ——…

Generate ASCII art using pyfiglet (571 fonts), cowsay, boxes, toilet, image-to-ascii, remote APIs (asciified, ascii.co.uk), and LLM fallback.

Hugging Face Hub CLI (hf) — search, download, and upload models and datasets, manage repos, query datasets with SQL, deploy inference endpoints, manage Spaces…

Search personal knowledge bases, notes, docs, and meeting transcripts locally using qmd — a hybrid retrieval engine with BM25, vector search, and LLM reranking.

Interactive active-memory ingestion skill for maintaining a persistent, compounding LLM knowledge base (the Memory Wiki).

Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart".

Designs production cloud architectures across all 6 pillars (AI/ML, Data Analytics, App Mod, Security, Multicloud, Sovereign Cloud).

从动效视频/截图反推出可直接喂给 LLM 还原动效的 Prompt 描述词。重点在「看懂运动」和「对齐你的意图」，不是搬运帧。触发词：分析动效、animate prompt、生成动效描述、动效 Prompt、animation analysis、帮我写这个动效的描述、还原这个动画

llama.cpp local GGUF inference + HF Hub model discovery.

Stop and consult this skill whenever your response would include specific facts about Anthropic's products.

当任务涉及 pyqcu/testing/qcu 或 quda 的单功能测试、PyTorch 参考或布局转换校验时使用。

Map where a claim, method, or design remains valid by defining variation axes, probing systematic or adversarial/boundary values, detecting breakpoints, and…

Select most-likely, least-likely, or otherwise inference-maximizing cases that can decisively support/refute a claim under explicit case-selection logic.

Select statistical inference/estimation method from design, distribution, sample size, pairing, multiplicity, and decision objective.

Update a cumulative evidence estimate/order as new studies or evidence blocks arrive and detect when inference changes materially over time.

Work on Sounio type checking and effect system (bidirectional inference, effect inference/enforcement, and type rules); use when editing…

Scan projects for credential leaks, secrets in code, insecure patterns, LLM API key exposure, PII leakage to external AI services, and .env/.gitignore…

Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart".

Collect and compare numerical results across hardware and software stacks for inference, SFT (full/LoRA) and RL.

Troubleshoots and debugs AWS Clean Rooms collaboration issues related to IAM roles, S3 bucket policies, KMS keys, Lake Formation permissions, and CloudWatch…

Orchestration d'un dialogue structuré entre deux LLMs (A et B) autour d'un même sujet, avec cadrage d'intention, génération d'angles morts, recherche…

Build and deploy an inference.sh flow — chain multiple apps into a pipeline with wired inputs/outputs.