176 件 ・ 関連度順
概要と使いどころ
Conducts privacy auditing of AI models including training data extraction testing, membership inference attacks, model inversion testing, and attribute inference assessment. Uses ML Privacy Meter and related tools to quantify privacy leakage. Keywords: model audit, membership inference, privacy meter, model inversion, training data extraction.
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。
Instruments code to track the flow of untrusted or sensitive data at runtime, enabling detection of injection vulnerabilities, data leaks, and privilege violations. Use when users need to: (1) Track untrusted input propagation through code, (2) Detect SQL injection, XSS, or command injection vulnerabilities, (3) Identify sensitive data leaks, (4) Monitor privilege escalation paths, (5) Perform dynamic taint analysis for security testing. Supports Python, Java, JavaScript, and C/C++ with configurable taint sources and sinks.
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。
Audit existing model training code — find reproducibility issues, data leakage, and missing best practices. Use when asked to "audit our training code", "is our training reproducible", or "check for training data leakage".
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。
Supabase security expert to audit RLS (Row Level Security), RBAC, relational databases, prevent data leakage, and utilize Supabase Linter / Ahli keamanan Supabase untuk audit RLS (Row Level Security), RBAC, database relasional, pencegahan kebocoran data, dan pemanfaatan Supabase Linter.
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。
Firebase security expert to audit Security Rules (Firestore/Realtime Database/Storage), authentication, API keys, data leakage prevention, and App Check configuration / Ahli keamanan Firebase untuk audit Security Rules (Firestore/Realtime Database/Storage), autentikasi, API keys, pencegahan kebocoran data, dan konfigurasi App Check.
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。
Decide what to PERSIST in your data structures (graph nodes, DB tables, cache layouts) based on the QUERIES you must answer, not the DEFINITIONS that justify the data. Definitions tell you logical identity; storage strategy is independent. Materializing every definitional component leads to O(N×M) blow-ups that masquerade as memory leaks. Use when: (1) designing a new graph-node type, edge type, DB table, or cache layout; (2) about to name a `subgraph` / `comprises` / `members` / `contains` collection on a parent entity; (3) memory growth proportional to (entities × per-entity-elements); (4) "memory leak" debugging that survives multiple targeted patches — the leak may be storage shape, not retention; (5) pre-implementation sketch includes "for each X, store all reached nodes"; (6) research/spec doc says "X = ⟨A, B, C⟩" and you're tempted to make it three storage edges per X.
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 Therapeutics Data Commons through the PyTDC Python package for registry discovery, approved dataset access, task-aware splits (scaffold, cold-start, temporal, combination), evaluator metrics, benchmark groups, and bounded molecular-oracle workflows. Use this skill to find which TDC datasets exist for a therapeutic task, load them with a split that does not leak, score predictions with the task's own official metric rather than a generic one, and run benchmark groups reproducibly. Also trigger on PyTDC, Therapeutics Data Commons, tdc.single_pred, tdc.multi_pred, ADMET Benchmark Group, scaffold split, get_split, or molecular oracles such as GSK3B, JNK3 and DRD2.
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。
Monitors dark web forums, marketplaces, paste sites, and ransomware leak sites for mentions of organizational assets, leaked credentials, threatened attacks, and threat actor communications to provide early warning intelligence. Use when establishing dark web monitoring coverage, investigating specific data breach claims, or enriching incident investigations with dark web context. Activates for requests involving dark web OSINT, leak site monitoring, credential exposure, Recorded Future dark web, or Tor hidden service intelligence.
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。
- Check and interpret data-breach exposure for an email, username, phone or name using Have I Been Pwned, the Pwned Passwords k-anonymity range API, DeHashed, IntelX and Snusbase. Use when checking breach or leak exposure, finding which services an account was registered with, interpreting a combolist or credential dump, assessing credential compromise, or auditing your own leaked personal data
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。
機械学習モデルを本番で使うために、データの仕様、再現可能な学習、品質評価、配備、監視、切り戻しを整理し、実装計画やレビュー項目にまとめるスキル。
Optimize Python code for reduced memory usage and improved memory efficiency. Use when asked to reduce memory footprint, fix memory leaks, optimize data structures for memory, handle large datasets efficiently, or diagnose memory issues. Covers object sizing, generator patterns, efficient data structures, and memory profiling strategies.
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。
Think and work like an expert AI Researcher. Use when a task calls for AI Researcher judgment. Reasons from data generating processes, inductive biases, and compute-data-algorithm trade-offs through train/val/test discipline, seed sweeps, ablation ladders, and standards like NeurIPS reproducibility checklists, model cards, and lm-eval-harness, while treating data leakage (Kapoor & Narayanan taxonomy), benchmark contamination, spurious correlations, and reward-model overoptimization as first-class failure modes.
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。
Expert at diagnosing and fixing performance bottlenecks across the stack. Covers Core Web Vitals, database optimization, caching strategies, bundle optimization, and performance monitoring. Knows when to measure vs optimize. Use when "slow page load, performance optimization, core web vitals, bundle size, lighthouse score, database slow, memory leak, optimize performance, speed up, reduce load time, performance, optimization, core-web-vitals, caching, profiling, bundle-size, database" mentioned.
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。
Check for data breaches and credential leaks using HaveIBeenPwned API and other sources. Use when: checking if an email or domain is in a known breach, assessing corporate credential exposure, password hygiene audits, or pre-engagement phishing simulation scoping.
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。
Audit feature engineering code for leakage, quality issues, and pipeline correctness. Use when asked to "audit our feature pipeline", "is there data leakage", or "find feature quality issues".
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。
Exploit AI applications using Indirect Prompt Injection. This skill focuses on hiding malicious instructions within data sources (web pages, documents, emails) that the LLM processes, causing the AI to execute unintended actions or leak data without direct user interaction.
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。
Run this skill for Phase 3 of the evidence verification pipeline — the adversarial audit step. Use when you need to check the evidence data lake for bad data before ingestion: dead links, wrong URL formats (tree/ vs blob/), subjective wording ("elite", "high-quality"), stale migration notes, benchmark catalog misuse/vendor-claim leakage, or skills whose star evidence conflicts with classified evidence level. Triggers on phrases like: "audit the data lake", "adversarial check", "ev-adversarial-audit", "check for noise in evidence", "flag bad evidence", "run the audit phase", "quality check the by-type files", or any reference to Phase 3 of the 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。
Reviews the instructions text and skills of a declarative agent, with its stated capabilities, knowledge sources and audience, for prompt-injection exposure, data-leakage paths, over-broad permissions and missing refusals, and returns an attack-surface map, a findings table with severity, quoted evidence and attack path, one pasteable fix per finding, a probe pack the owner can run and a refusal-coverage table. Use when the user asks to "red-team this agent", "check these instructions for prompt injection", "could this agent leak data", "review the permissions on my agent" or "what should this agent refuse". Do not use for checking a SKILL.md against the format rules, use skill-file-reviewer instead; do not use for designing the launch test set, use agent-evaluation-plan instead. Drafts for human review; never approves, authorises or signs 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。
Background processing and heavy computations using Dart Isolates with Isolate.run(), compute() function, and platform channels. Use this skill when performing CPU-intensive operations without blocking UI (image processing, data parsing, encryption), implementing background tasks that prevent jank or frame drops, handling large file operations (JSON parsing, database migrations), running parallel computations, fixing UI freezes caused by main thread blocking, using compute() for one-off background work, creating long-lived Isolates with ReceivePort/SendPort, implementing multi-threaded algorithms, debugging isolate communication issues, or preventing memory leaks from unclosed ReceivePorts. Covers isolate spawn, message passing, ReceivePort cleanup (CRITICAL for leak prevention), SendPort communication patterns, error handling, serialization rules, and performance optimization for compute-heavy operations.
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。
Rigorous methodology for evaluating ML models on established benchmarks. Covers proper train/val/test splits, baseline verification from original papers, exact metric formula discrepancies, data-leak detection checklist, multi-seed robustness, and honest reporting templates. Use when claiming to beat published baselines, writing methods papers, or auditing existing results.
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。
Validates predictive models on omics and biomedical data with nested cross-validation, group/batch/temporal-aware splits, the full data-leakage taxonomy, probability calibration, decision-curve net benefit, optimism correction, sample-size planning, and TRIPOD+AI reporting. Use when estimating model performance honestly, choosing a CV scheme, detecting leakage, or judging whether reported discrimination means the model is actually useful. For feature selection itself see machine-learning/biomarker-discovery; for confirmatory-trial inference see clinical-biostatistics/trial-reporting.
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。
ローカルで繰り返しているHermesの作業手順から個人情報や認証情報を除き、公開・再利用できるECCスキルやリリース資料に整理するスキル。
Performs bounded, local exploratory analysis of explicitly supported scientific files. Supports redacted CSV/TSV/JSON profiles; optional NumPy, HDF5, FASTA/FASTQ, and basic image metadata inspection; missingness/leakage audits; outlier and transformation sensitivity; and rigorous EDA report scaffolds. Other domain formats are reference-only and unknown formats fail closed.
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。
Audit and harden a Convex app's authorization: identity-from-arg impersonation, missing per-document ownership checks, public queries leaking data by a client-supplied id, and writes into a parent/container the caller doesn't own. Scans for the 4 shapes, applies requireIdentity/requireOwner, verifies with tsc. TRIGGER on 'secure my app', 'audit auth/authz', 'who can access this data'. SKIP when there is no convex/ directory.
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。