Research & Data

Design a domain ontology in Neo4j with a user, optionally loading a folder of PDFs and extracting entities against that ontology. Use whenever a user wants to design a domain model / ontology, turn documents into a queryable knowledge graph, do legal / HR / finance / regulatory document analysis, structure unstructured text, or asks anything like 'help me model my domain', 'how do I make a graph from my documents', 'I have a folder of contracts/CVs/reports — help me explore them', 'extract entities from PDFs', 'build a domain model' — even if they don't say 'ontology' or 'knowledge graph' explicitly. Always trigger when the conversation involves domain modeling, PDF documents plus Neo4j, or…

Country-specific business research and market-entry planning. Asks which country and which business, then researches competitors, valuations, stock market and GDP context, economic stability, local networks and key people, and builds a project board with a dated action plan, optional meeting scheduling, optional alerts, and an optional recurring refresh. Use whenever the user wants to start, expand, or bring a business or business concept into a specific country, asks "does [country] need [business]", wants competitor or market research for a country, wants to find contacts or networks for a venture abroad, or wants to know how to register a company there, even if they never say "market…

Security gate for shipping and running products at scale. Finds exposed API keys and secrets (source, git history, frontend/mobile bundles, Dockerfiles, CI workflows, coding-agent configs), reviews AI gateway / LLM, payment and analytics integrations, flags single points of failure, and produces a go/no-go report against a fixed checklist. Use this skill whenever the user is about to deploy, release, launch, go live or cut a production build; adds or changes an integration such as OpenAI, Anthropic, Bedrock, Vertex, an AI gateway, MCP server, Stripe, Razorpay, Adyen, PayPal, Segment, PostHog, Mixpanel, Sentry or Google Analytics; asks for a security audit, pre-launch review, hardening pass…

Use when reviewing, auditing, threat-modeling or planning the security of a web app, API, dashboard, MCP server, automation (n8n), self-hosted service, Docker host or homelab; when deciding how much security a project needs; when hunting vulnerabilities (BOLA/IDOR, broken RLS, leaked secrets, exposed admin UIs, outdated images or packages); before a launch or deploy; after a suspected breach; or when writing AGENTS.md/CLAUDE.md rules for an AI-built project.

Produce a graded daily (or weekly, or monthly) health and security report for a website — traffic and errors from the site's own edge analytics, SEO/AEO checker scores, and code-host scanning (dependencies, SAST, secrets) — with evidence behind every figure, a pre-send verification gate, and owner-approved sending. Use when asked to run, set up, or continue a site's health/security report on any stack (Cloudflare, AWS CloudFront, Azure, Fastly, Akamai; GitHub, GitLab, Bitbucket, Azure DevOps, Gitea). Always start with the stack intake in references/stack-intake.md; never assume a vendor.

Programmatically add or edit visuals, pages, and visual properties inside a Power BI Desktop .pbix file (PBIR Enhanced Report Format) without corrupting it. Use when the user asks to build, modify, generate, or automate a Power BI dashboard directly in a saved .pbix file — triggers include "edit pbix", "add visuals to Power BI", "build a Power BI dashboard programmatically", "modify Report/definition/pages", "patch a pbix", "generate Power BI report from a dataset", or anywhere a .pbix file is the input and output. Also use when the user complains that a hand-edited .pbix opens as "corrupted" (Mashup / SecurityBindings errors) — this skill explains the root cause and the fix.

Audit local SEO, including Google Business Profile, NAP consistency, citations, reviews, local schema, location pages, and multi-location structure. Use when the user asks about local SEO, a local business, the map pack, or a GBP profile; use seo-maps for geo-grid rank tracking.

Evaluate survey results like a survey statistician - cleaning, distributions, cross-tabs, significance, weighting, subgroup reports, anonymity, management summary. Statistical only.

Solve a fluid-flow problem end to end with OpenFOAM and produce a verified HTML report (problem, assumptions, theory, decisions, pre-flight test, simulation, comparison, insights). Use when given a CFD problem statement in cases/<name>/problem.yaml or asked to simulate a flow and report on it.

Create a senior-level system design (HLD) guide for any product or company: requirements, capacity, architecture & ER diagrams, schema, APIs and trade-offs, in light and dark themes.

Run when the user says "cite viewer" or "build the viewer": a local click-through page where each citation in a draft opens the cited passage from the source, with optional links to the record.

Reddit research via the Arctic Shift archive API — subreddit analysis, engagement metrics, user activity, deleted/removed post recovery, tone assessment, rules, and topic discovery. Direct reddit.com access is blocked in this environment, so the skill also says plainly which questions cannot be answered rather than substituting web search. Use when the user asks what Reddit says about a topic, wants a subreddit or user analyzed, needs sentiment or demand signal from Reddit threads, or asks to recover a deleted/removed post.

Convert PDFs such as textbooks, lecture slides, technical manuals, and research papers into detailed, well-structured study notes with automatically selected diagrams, tables, formulas, and clear explanations. Use when a user uploads a PDF and asks for notes, study material, or a structured explanation.

Search the Metropolitan Museum of Art collection, research artworks, inspect visual references, and download artwork images to local folders. Use for Met collection research, broad art discovery, and reference gathering.

業界別・希望職種別・選考フェーズ（ES・1次面接・2次面接・最終面接・オファー比較）別・新卒/中途別のパターンから選択して、一次情報（有価証券報告書・決算説明資料・中期経営計画・統合報告書・採用ピッチ資料）と参考情報としての口コミに基づく徹底的な企業分析と、競合差別化・約1分の自己紹介・志望動機・仮説検証型逆質問・想定質問100問の作成を行うスキル。ユーザーが企業分析、企業研究、業界・競合比較、IR分析、または特定企業の選考・面接対策を求めたときに使用する。

Build, review and fix evals for LLM apps, RAG pipelines and agents the way the aievals.co guide teaches. Error analysis on real traces before any metric, code checks before LLM judges, one binary PASS/FAIL judge per failure mode, judges calibrated against human labels (PASS and FAIL recall, not agreement), train/dev/test splits, regression sets, Wilson intervals, and paired release gates in CI. Use when asked to add evals or tests to an LLM feature, write or debug an LLM-as-judge prompt, label or review traces, pick metrics or sample sizes, compare two prompts or models, check whether an existing eval can be trusted, or wire evals into CI.

Turn a raw brief, RFP, discovery transcript, or feature spec into an architect-grade analysis — the architecturally significant requirements pulled out and split into functional / quality-attribute / constraint, each sharpened into something testable, prioritised by business value and architectural impact, checked for requirements that quietly contradict each other, and turned into measurable quality scenarios with the design strategies that would satisfy them. Use whenever a brief, pre-sales input, RFP, discovery-call notes, kickoff document, or set of requirements needs to be assessed before estimation or design — and whenever someone asks what is architecturally significant here, what...

Use for keyword research: turning a seed keyword, business type or topic into an intent-clustered keyword map with real search volumes, AI search volume and the fan-out questions a page must answer. Step 1 of the 5-skill SEO system. Also use when asked what to write about, which keyword to target, or to find keywords for a page.

Produce and review software architecture documentation grounded in open standards — ISO/IEC/IEEE 42010, the arc42 template, the C4 model, the 4+1 view model, and Architecture Decision Records. Use when you need to document a system's architecture, decide which views to write for which stakeholders, structure an architecture description, capture decisions with their rationale, or review existing architecture docs for completeness and clarity.

Use when planning, doing, or writing scientific research.

Run /audit on a greenfield project, an existing codebase with missing docs, or one area (/audit src/auth) to bootstrap the project's AI context, the AGENTS.md files every later skill reads. Writes tool agnostic AGENTS.md plus thin CLAUDE.md pointers, adding only what is missing; never overwrites curated content.

Audits a website or vibe-coded app (Lovable, Bolt, v0, Replit, Next.js, React, plain HTML) for tracking that loads or sends data BEFORE the visitor agrees in the cookie banner - ad pixels (Meta, Google, TikTok, LinkedIn), Google Tag Manager, analytics, session replay, chat widgets, embeds, hosting-injected analytics, and customer emails or IDs sent to trackers. Read-only - a code scan plus a live fresh-visit network check - ending in an inventory and a minimal fix list. Use when someone asks "is

Audit a document claim by claim: recompute every number, read charts back against the text, and check that cited sources exist. Use when the user asks to audit, verify, fact-check or sanity-check a report, PDF, deck, spreadsheet or draft; asks whether the numbers add up; or is about to send a document that contains figures, charts or citations. Verdicts are computed by a bundled script, not by the model.

Write a survey whose every question feeds a decision, with the sample size and analysis worked out first.