データ・AIのスキル

AI and ML expert covering PyTorch, TensorFlow, Hugging Face, scikit-learn, LLM integration, RAG pipelines, MLOps, and production ML systems

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。

LLM-as-judge evaluation framework with 5-dimension rubric (accuracy, groundedness, coherence, completeness, helpfulness) for scoring AI-generated content quality with weighted composite scores and evidence citations

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。

Research-backed workflow to refresh existing agent prompts/frontmatter with diff-based risk scoring, TDD gates, and ecosystem validation.

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。

Molecular featurization hub with one consistent interface over 100+ featurizers. Fingerprints (ECFP/Morgan, MACCS, atom pair, topological torsion, Avalon, RDKit, ERG), RDKit and Mordred descriptor sets, pharmacophore and 3D shape descriptors, scaffold keys, and pretrained embeddings (ChemBERTa, ChemGPT, MolT5, GIN, Graphormer) through a common transformer API with caching and parallelism. Use this skill to convert SMILES into model-ready feature matrices for QSAR, virtual screening, and molecular ML, and to choose between featurizer families. Also trigger on molfeat, MoleculeTransformer, FPVecTransformer, PretrainedHFTransformer, molfeat model store, or featurizer selection.

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。

Molecular ML with diverse featurizers and pre-built datasets. Use for property prediction (ADMET, toxicity) with traditional ML or GNNs when you want extensive featurization options and MoleculeNet benchmarks. Best for quick experiments with pre-trained models, diverse molecular representations. For ready-made ADMET numbers without training a model use admet-prediction; for benchmark datasets and task-aware splits use pytdc.

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。

DiffDock and DiffDock-L diffusion-based molecular docking. Use for blind protein-small-molecule pose prediction from a PDB file or sequence plus SMILES/SDF/MOL2, batch docking over a CSV of complexes, virtual screening triage, sampling multiple poses per complex, and reading the confidence score correctly. Also trigger on DiffDock, DiffDock-L, inference.py, confidence_model, samples_per_complex, ESM embedding preparation for docking, or blind docking without a defined box. Not for binding affinity prediction — the confidence score ranks pose plausibility, not potency.

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。

Protein language models through the EvolutionaryScale `esm` Python SDK. Generate and embed sequences with ESM3 (multimodal sequence, structure and function prompting), extract per-residue and mean-pooled embeddings with ESM C, fold sequences with ESMFold2, and run inference locally or against the Forge and Biohub hosted clients. Use this skill for protein representation learning, variant effect and mutational scanning from likelihoods, sequence generation and inpainting, structure prediction from sequence alone, and embedding features for downstream models. Also trigger on esm, ESM3, ESMC, ESM Cambrian, ESMFold2, `from esm.models`, ESMProtein, GenerationConfig, forge.evolutionaryscale.ai, biohub.ai, or ESM_API_KEY.

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。

Query the ChEMBL database web services for measured bioactivity data, compound records and calculated properties, targets, assays, mechanisms of action, drug indications and warnings. Use this skill to build curated SAR or QSAR datasets for a target, look compounds up by SMILES, InChIKey, name, or ChEMBL id, run similarity and substructure searches, and check what chemistry is already known against a protein. Also trigger when a query mentions ChEMBL ids (CHEMBL...), pChEMBL values, IC50/Ki/Kd/EC50 retrieval, assay confidence scores, or ebi.ac.uk/chembl.

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。

Pure-Python knowledge base statistics dashboard. Reads shelf-index and log.md; emits Inventory, Layer distribution, Domain distribution, Recent Activity, and Staleness sections. No agent dispatch. Read-only.

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。

Manage the project's layer vocabulary — list allowed layers with file counts, add new layers, or safely remove unused layers. Reads the shelf-index for usage data; writes CLAUDE.md atomically. No agent dispatch.

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。

Health-check the project knowledge base. Looks for contradictions between files, stale claims that newer sources have superseded, orphan files with no inbound cross-references, important concepts mentioned but lacking their own page, missing cross-references, and data gaps. Returns a structured report; does not auto-fix.

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。

Strip AI writing patterns from prose across five registers — casual, professional, academic, technical, and narrative. Applies 200+ pattern rules to eliminate predictable LLM tells. Triggers on clean up writing, remove AI slop, edit prose, make this sound less AI.

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 a repository's type system and generate type-safe DAG execution pipelines with GraphSentry-style certificate verification. This skill should be used when building LLM-driven workflows that need deterministic type safety, when composing typed operations into validated execution graphs, or when adding contract-checked pipeline orchestration to any codebase. Supports Python (Pydantic) and TypeScript (Zod).

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。

Data analysis workflow kit using the ALIVE loop (Ask, Look, Investigate, Voice, Evolve)

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。

Package codebases into LLM-optimized markdown, and run the Deep Think Authority pipeline (context-builder → Deep Think review → structured AUTHORITY → Flash/Antigravity implementation agent). Use when generating project context, preparing Deep Think reviews, or binding a weaker coding agent to a senior reasoning pass.

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 and record evidence-backed LLM and AI-search citation checks across approved assistant/search surfaces with timestamped queries, locales, modes, cited URLs, snippets, screenshots or transcripts, and caveats. Use this skill when the user asks whether ChatGPT, Perplexity, Claude, Gemini, Copilot, or another assistant cites a target site, but do not claim citation without direct evidence.

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。

Design human-visible, machine-readable website layers for AI assistants, RAG, and agentic search. Use this skill when the user asks for llms.txt, LLM discovery plans, visible TLDR or direct-answer blocks, source trails, entity pages, markdown alternates, AI crawler policy, topic-to-URL maps, assistant-readable public content, or LLM-friendly architecture that must preserve canonical SEO strategy without hidden bot-only facts or duplicate story bodies.

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。

Build evidence-labeled semantic cores for SEO, LLM-readable architecture, and site information design. Use this skill when the user asks for semantic core, query clusters, search intents, audience segments, entity/topic maps, language or locale priorities, competitor/source evidence, pillar-topic planning, or a structured handoff before URL architecture and internal linking. It must mark unknown search volume, difficulty, and ranking data as unknown instead of inventing metrics.

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。

Specialized agentic and context engineering prompt methodology for Claude. Use when designing AI agents — system prompts, tool definitions, context window architectures, failure mode planning, multi-agent coordination, agent evaluation. Trigger on: "design an agent," "build a system prompt for an agent," "architect an agent's context," "design tool definitions," "agent system prompt," "context window architecture," "tool interface design," "evaluate agent behavior," "multi-agent coordination," "context engineering for agents," "ACI design." Provides 12 tested approaches for agent architecture (designing agents, not using them). Do NOT use for evaluating or scoring individual non-agent prompts (use rootnode-prompt-validation if available). Do NOT use for general software architecture that hosts agents (use standard technical reasoning for that).

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。

Eight tested identity approaches for Claude prompts, each shaping depth, vocabulary, reasoning style, and what Claude treats as obvious vs. requiring explanation. Use when the user wants a specific identity template — retrieving, reviewing, customizing, or building a role definition. Trigger on: "give me the Strategic Advisor identity," "show me the identity template for," "I need a Technical Architect role," "build a custom role for," "show me all available identities." Also use when reviewing a prompt's identity layer or when output lacks domain-appropriate depth. Provides 8 tested approaches across strategy, technical, research, communications, and operations domains plus a template for building custom identities. If the user is unsure which identity fits their task, use rootnode-block-selection first if available. Do NOT use for evaluating complete prompts — use rootnode-prompt-validation if available. Do NOT use for full prompt assembly — use rootnode-prompt-compilation if available.

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。

Evaluates and scores Claude prompts and system prompts using a six-dimension Scorecard with anchored 1-5 rubrics. Trigger on: "review my prompt," "score this prompt," "rate my prompt," "evaluate this prompt," "improve my prompt," "what's wrong with my prompt." Also trigger on: "why isn't my prompt working," "my prompt produces bad output," "help me fix this prompt," "is this prompt any good." Scorecard covers Objective Clarity, Context Specificity, Reasoning Fit, Output Precision, Behavioral Calibration, and Architectural Efficiency, plus a five-question Output Evaluation Rubric for assessing actual output. Also use when the user pastes a system prompt and asks for feedback. Activate whenever structured quality assessment of a single prompt is the primary need. Do NOT use for project-level audits involving Custom Instructions architecture, knowledge file organization, or multi-file Project structure (use rootnode-project-audit if available).

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。

Tested output format specifications for Claude prompts. Use when the user wants a specific deliverable structure — retrieving, reviewing, customizing, or building output formatting. Trigger on: "give me the Executive Brief format," "show me the output template for," "I need the Decision Matrix structure," "show me available output formats," "build a custom output format for." Provides 10 ready-to-use structures (executive briefs, technical designs, research summaries, implementation plans, decision matrices, competitive analyses, post-mortems, stakeholder updates, strategic memos, process documentation) plus a custom template, with per-section length guidance and format constraints tested against Claude formatting defaults. If unsure which format fits, use rootnode-block-selection first if available. Do NOT use for evaluating existing prompts (use rootnode-prompt-validation if available) or choosing reasoning/identity approaches (use rootnode-block-selection if available).

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。

Builds complete, optimized Claude prompts and Project scaffolds from task descriptions. Use when building, designing, creating, or structuring a Claude prompt or system prompt. Trigger on: "build me a prompt," "compile a prompt for," "create a system prompt," "write a Claude prompt," "design a prompt for," "help me write a prompt," "scaffold a Claude Project," "design a Project for." Uses a four-stage methodology (Parse, Select, Construct, Validate) and a tested 5-layer architecture (Identity, Objective, Context, Reasoning, Output + Quality Control). In Project Mode, the Compiler is global-layer-aware — checks User Preferences, installed Skills, and MCP Connectors to avoid redundancy and flag gaps. Do NOT use for auditing existing prompts (use rootnode-prompt-validation if available) or diagnosing existing Projects (use rootnode-project-audit if available). Run on Opus 5 or Sonnet 5 at `high` effort (both defaults); depth reduces on legacy models.

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。