「ai safety」の検索結果

常時稼働するAIエージェントの起動・停止、ログ監視、権限管理、段階的な更新と切り戻しを整理し、障害対応や成功率・コストの評価を支えるスキル。

• 常時稼働エージェントの運用設計
• 失敗が増えた際の障害対応
• 更新と切り戻しの手順整備

Search and retrieve arXiv academic papers by topic, category, or paper ID — with AlphaXiv-enriched AI-generated overviews. Uses arXiv Atom API across cs.AI/cs.LG/cs.CL/cs.CR/cs.MA/cs.SE/cs.IR. Three workflows: Latest, Search, Paper. USE WHEN arxiv, papers, latest papers, research papers, recent ML papers, paper lookup, summarize paper, latest LLM papers, AI safety papers, cs.AI latest. NOT FOR general research (Research), URL parsing, or annual reports.

日本語の概要は準備中です。原文の説明を表示しています。

Extract structured data from LLM responses with Pydantic validation, retry failed extractions automatically, parse complex JSON with type safety, and stream partial results with Instructor - battle-tested structured output library

日本語の概要は準備中です。原文の説明を表示しています。

Writes, reviews, and debugs idiomatic Rust code with memory safety and zero-cost abstractions. Implements ownership patterns, manages lifetimes, designs trait hierarchies, builds async applications with tokio, and structures error handling with Result/Option. Use when building Rust applications, solving ownership or borrowing issues, designing trait-based APIs, implementing async/await concurrency, creating FFI bindings, or optimizing for performance and memory safety. Invoke for Rust, Cargo, ownership, borrowing, lifetimes, async Rust, tokio, zero-cost abstractions, memory safety, systems programming.

日本語の概要は準備中です。原文の説明を表示しています。

Analyze text and images for harmful content using Azure AI Content Safety (@azure-rest/ai-content-safety). Use when moderating user-generated content, detecting hate speech, violence, sexual content, or self-harm, or managing custom blocklists.

日本語の概要は準備中です。原文の説明を表示しています。

Constitutional AI and safety guardrail prompts for aligned LLM behavior

日本語の概要は準備中です。原文の説明を表示しています。

Conduct a structured ethical review of an AI or ML feature, model, or product. Use when preparing to deploy an AI system, assessing algorithmic risk, auditing a model for bias, or producing a responsible AI impact assessment. Produces a structured ethics review covering fairness, transparency, privacy, safety, accountability, and societal impact with a risk tier score, pre-deployment checklist, and prioritised mitigations.

日本語の概要は準備中です。原文の説明を表示しています。

Audits legal AI skills, prompts, workflows, MCP/tool instructions, and agent packages for prompt injection, malicious instructions, unsafe scripts, suspicious metadata, credential exposure, exfiltration paths, persistence mechanisms, and supply-chain risk. Use before installing, importing, publishing, promoting, modifying, or trusting third-party or generated AI skills and legal workflows. Do not use as a generic code review unless skill trust, workflow safety, prompt injection, or supply-chain hygiene is in scope.

日本語の概要は準備中です。原文の説明を表示しています。

This skill channels the strategic and scientific reasoning of Demis Hassabis, CEO and co-founder of Google DeepMind, AlphaGo and AlphaFold, and 2024 Nobel Prize in Chemistry. Use this skill whenever you are evaluating AI for scientific discovery, tackling "root node" problems, designing reinforcement learning systems, or discussing AGI timelines, safety, and global governance. Reach for it when the user faces massive combinatorial search spaces, wants to apply AI to physical/biological sciences (like digital biology), or needs to balance rapid AI scaling with the rigorous scientific method. Apply these mental models to shift the focus from building consumer apps to using AI as the ultimate meta-solution for understanding reality.

日本語の概要は準備中です。原文の説明を表示しています。

Automatically repair code violations of temporal properties using model-checking counterexamples as guidance. Use when working with formal verification results, temporal logic property violations (LTL, CTL), model checking counterexamples, or when asked to repair property violations, fix counterexamples, repair temporal properties, fix model checking violations, or repair code based on counterexamples. Applicable to concurrent systems, state machines, synchronization issues, safety/liveness properties, and resource management violations.

日本語の概要は準備中です。原文の説明を表示しています。

当需要为 AI 定义工具接口、设计调用规范、实现工具发现与编排机制时调用此 skill。典型场景包括：设计 AI agent 的工具集、定义 JSON Schema/XML/TypeScript 格式的工具描述、实现工具权限控制与并行调度、设计子代理委托架构。 不适用于：定义 AI 身份（persona-design）、安全规则（safety-guardrails）、记忆存储（memory-system）。 关键 trigger 信号：AI 需要调用外部 API 或服务、工具数量超过 10 个需要发现机制、需要区分工具权限层级、需要子代理并行执行、工具调用涉及敏感操作需要审批。

日本語の概要は準備中です。原文の説明を表示しています。

Validate email addresses for deliverability, check reputation scores, verify MX records, and assess email-sending safety. Covers deliverability checks, reputation scoring, bounce risk assessment, and bulk validation for outreach campaign preparation.

日本語の概要は準備中です。原文の説明を表示しています。

Provider-independent VFX compositing direction for generated video, ads, trailers, product films, social clips, explainers, mixed-source edits, and image/video composites. Use when an agent must plan, brief, generate, integrate, repair, or quality-check visual effects composites involving plates, alpha/mattes, keying, roto, tracking, stabilization, matchmove, lighting/color integration, grain/noise/sharpness, shadows/reflections/contact, depth/atmosphere, AI-generated elements, clean plates, editor/colorist handoff, rights/safety, or compositing QA.

日本語の概要は準備中です。原文の説明を表示しています。

Generate, edit, composite, stream, and production-review still images with OpenAI GPT Image models. Use when an agent must choose between GPT Image 2 and legacy GPT Image/DALL-E integrations, select the Image API or Responses API, build prompts and reference-image workflows, use masks or multiple inputs, preserve identity or brand details, render in-image text, handle output files and partial images, estimate cost and rate limits, migrate deprecated image code, or apply OpenAI image safety, consent, privacy, provenance, and rights requirements. Do not use for OpenAI video or audio generation.

日本語の概要は準備中です。原文の説明を表示しています。

Azure AI Content Safety REST SDK for TypeScript workflow skill. Use this skill when the user needs Analyze text and images for harmful content with customizable blocklists and the operator should preserve the upstream workflow, copied support files, and provenance before merging or handing off.

日本語の概要は準備中です。原文の説明を表示しています。

Agent-Friendly CLI Spec v0.1 workflow skill. Use this skill when the user needs Design spec with 98 rules for building CLI tools that AI agents can safely use. Covers structured JSON output, error handling, input contracts, safety guardrails, exit codes, and agent self-description and the operator should preserve the upstream workflow, copied support files, and provenance before merging or handing off.

日本語の概要は準備中です。原文の説明を表示しています。

This skill channels the strategic and scientific reasoning of Demis Hassabis, CEO and co-founder of Google DeepMind, AlphaGo and AlphaFold, and 2024 Nobel Prize in Chemistry. Use this skill whenever you are evaluating AI for scientific discovery, tackling "root node" problems, designing reinforcement learning systems, or discussing AGI timelines, safety, and global governance. Reach for it when the user faces massive combinatorial search spaces, wants to apply AI to physical/biological sciences (like digital biology), or needs to balance rapid AI scaling with the rigorous scientific method. Apply these mental models to shift the focus from building consumer apps to using AI as the ultimate meta-solution for understanding reality.

日本語の概要は準備中です。原文の説明を表示しています。

Prioritize drug targets from a ranked gene list (e.g., scRNA-seq DE output) by orchestrating parallel API queries against UniProt, OpenTargets (with integrated DepMap CRISPR essentiality + gnomAD constraint), PubMed, the Human Protein Atlas (HPA), and ChEMBL tool compounds, then re-ranking by a composite score combining protein localization, druggability, disease genetics, tissue specificity (safety), focus-cell-type expression, CRISPR essentiality, LoF safety constraint, and research maturity. Use whenever the user wants to filter, triage, prioritize, or "do due diligence" on a list of candidate genes for drug discovery, especially after a DE / DEG analysis when they say things like "which of these should I follow up on", "filter for druggable targets", "make a target dossier", "rank these for tractability", "annotate these genes for druggability", or "build a target report". Trigger even when the user says just "filter these candidate genes" or hands over a CSV from a DE pipeline.

日本語の概要は準備中です。原文の説明を表示しています。

Traditional Chinese internal arts — tai chi chuan and qigong — as transmitted by classical lineages (Chen, Yang, Wu, Wu/Hao, Sun for tai chi; Ba Duan Jin, Yi Jin Jing, Liu Zi Jue, Wu Qin Xi, and other named sets for qigong) with an emphasis on alignment, rooted stance, breath-posture-mind coordination (the three harmonies), and the traditional lineage voice that Dr. Yang Jwing-Ming's YMAA line preserves. Also covers the safety boundary between traditional practice and clinical "medical qigong" claims. Use for queries about tai chi form learning, qigong set selection, internal-arts pedagogy, and the practical question of where to start.

日本語の概要は準備中です。原文の説明を表示しています。

Plan a hiking tour with trail selection by difficulty (SAC/UIAA), time estimation using Munter's formula, elevation analysis, and safety assessment. Covers multi-day hut-to-hut tours, day hikes, and alpine routes with terrain classification and group fitness considerations. Use when planning a day hike or multi-day trekking tour, selecting trails appropriate for a group's fitness and experience, estimating realistic hiking times, or planning hut-to-hut tours with overnight logistics.

日本語の概要は準備中です。原文の説明を表示しています。

Sets up safe working habits and a change log so you always know what Cowork has done to your files — and can undo anything. Use when setting up file protection guardrails, for establishing safe automation habits, during initial access configuration, when granting file permissions for the first time, or when the user says "is cowork safe", "file safety", "what if it deletes my files", "audit trail", "safety setup", "I'm nervous about file access", "how do I trust cowork with my files", "can cowork mess up my computer", "I heard someone lost files", or "how do I stay in control".

日本語の概要は準備中です。原文の説明を表示しています。

Use when building autonomous LLM agents operating in live environments. Protects against prompt injection, tool abuse, data exfiltration, and runaway loops. Keywords: guardrails, prompt injection, IPI, safety, sandboxing, NeMo, LLM Guard.

日本語の概要は準備中です。原文の説明を表示しています。

Use when a task needs the judgment of a Commercial Diver — planning a surface-supplied or SCUBA-mode dive against no-decompression limits, sizing a bailout/reserve gas supply before descent, choosing wet vs. dry (habitat) underwater welding for a repair, scoping an underwater NDT inspection, or deciding whether to abort a dive against a safety threshold.

日本語の概要は準備中です。原文の説明を表示しています。

Use when a task needs the judgment of a bicycle repairer — diagnosing drivetrain skip and quoting chain-vs-cassette replacement, torquing carbon components to spec, bleeding hydraulic disc brakes, truing a wheel to tension (not just by eye), or deciding whether a safety-critical fastener issue needs to be declined-in-writing rather than patched.

日本語の概要は準備中です。原文の説明を表示しています。