Skills
96,770 packaged capabilities and playbooks for agents. Install any of them with one command.
Productivity
Researches hotels, flights, attractions, short-term rentals, and live events via the Crawlora API — Booking.com, Expedia, Agoda, TripAdvisor, Trip.com, Airbnb, and Ticketmaster — returning clean JSON. Use when the user wants to search or compare hotel/stay prices and reviews, look up flight options, find attractions/things-to-do or concerts/events, or research an Airbnb host or listing.
UiPath automation discovery — mines Slack/email/wikis/CRM/HRIS/ERP for repetitive work, SPOFs, and replicable models; produces a 4-tier prioritized opportunity report with UiPath implementation paths, then sizes build effort (complexity band → pack-hours → contingency). Use to discover automation opportunities, find what to automate, estimate/size a delivery, or run an internal automation audit across an organization. For building a specific automation→uipath-rpa, authoring a Flow→uipath-maestro-flow, working with agents→uipath-agents.
Review applications and verify task submissions on OpenAnt. Use when the agent (as task creator) needs to review applicants, accept or reject applications, approve or reject submitted work, download submission files, or give feedback on deliverables. Covers "review applications", "approve submission", "reject work", "check applicants", "verify task", "download submission files".
Send WhatsApp messages to other people or search/sync WhatsApp history via the wacli CLI (not for normal user chats). Use when the user asks to send a WhatsApp message, text someone on WhatsApp, search WhatsApp chat history, sync WhatsApp conversations, backfill message history, or forward a file via WhatsApp to a third party.
Search for financial reports using Exa advanced search. Near-full filter support for finding SEC filings, earnings reports, and financial documents. Use when searching for 10-K filings, quarterly earnings, or annual reports.
Keyless, self-hostable multi-engine web search and clean-Markdown page reader for AI agents. Use it when the user asks to search the web, look something up online, find current or recent information, research a topic, fetch or read a URL, find academic papers or GitHub repositories, verify a claim against live sources, or configure and verify its NordVPN egress proxy. Fuses many keyless engines via the ddgs metasearch (plus a self-hosted SearXNG it can start itself), then extracts pages to clean Markdown, fenced as untrusted and paginated. Commands: init (bring it online), web-search (find pages), web-fetch (read a URL), web-open (page through a fetched document), arxiv (search papers), github (search repositories).
Searches academic papers on arXiv, analyzes research content, builds literature reviews, and generates citations for academic writing. Triggers when users need to find, understand, or cite arXiv papers, extract TikZ figures from LaTeX sources, or explore citation networks and coauthor relationships.
Query the bundled research knowledge graph for methodology guidance. Routes questions through a 3-tier knowledge base — WHY (research claims), HOW (guidance docs), WHAT IT LOOKS LIKE (domain examples) — plus structured reference documents. Returns research-backed answers grounded in specific claims with practical application to the user's system. Triggers on "/ask", "/ask [question]", "why does my system...", "how should I...".
ACTIVATE when preparing for board meetings, creating board decks, drafting board materials, developing executive summaries for directors, writing talking points, crafting board agendas, addressing governance matters, preparing for investor questions, or managing board communication. Critical for any board-facing materials, shareholder communication, or governance compliance.
Extract core methodologies from books and generate the corresponding skill. Use this skill when the user says "extract the methodology from 《XXX》", "generate a skill for 《XXX》", "turn this book's methodology into a skill", or has a similar request. Suitable for methodology extraction and skill generation from books of various kinds, including business, management, personal growth, and technical books.
Guides management consulting-style work—engagement framing, hypothesis-driven problem structuring, issue trees, business cases, operating model and capability design, strategic options analysis, workshop facilitation, and executive recommendations (not legal advice). Use when diagnosing a business problem, structuring a strategy or transformation initiative, building a business case for leadership, designing target operating models, preparing steerCo or board recommendations, or advising on build-vs-buy and portfolio priorities—not for detailed requirements/BRDs (business-analyst), multi-team delivery tracking (technical-program-manager), contract negotiation (commercial-counsel), revenue accounting (senior-revenue-accountant), applied AI architecture (applied-ai-architect-commercial-enterprise), or system ADRs (senior-system-architecture). Canvas/TAM: business-model-researcher. Comms: communication-lead. M&A closing: transaction-manager. M&A principal/IC: transaction-principal.
PRODUCES a citation-budget plan/report (`output/CITATION_BUDGET_REPORT.md`); does NOT edit the draft. Allocates an in-scope per-H3 budget (NO NEW FACTS) so drafts stop failing the global unique-citation gate; citation-injector applies it.
Analyzes professions/jobs for Claude Code automation opportunities. Triggers: 'how can Claude help me', 'what can Claude do for', 'I'm a [profession]', 'help me as a [job]', 'I work as', describing their work + asking about Claude. Use whenever user mentions their profession/role and wants to discover what Claude can automate.
Generate a graded exercise set — easy to advanced — that lets a learner verify whether they actually understand a topic, concept, method, derivation, or paper, with solutions kept in a separate file for honest self-testing. Use whenever the user says "give me exercises", "quiz me on X", "test whether I actually get this", "problems from easy to hard", "practice questions with solutions", "how do I check if I understood this", or wants a self-check after learning something. Produces two artifacts: an `exercises.md` (problems only, tagged by difficulty tier and concept) and a `solutions.md` (worked answers plus a self-check rubric and a pointer to what to revisit on a miss). This skill builds the verification material; it does not teach the concept (use professor-mentor-technical-teaching for that) and it does not sequence a learning path (use flow-learn-new-topic for that). It is one of the assessment steps those flows hand off to at an exit gate (the other being a discussion round).
Control the user's Chrome browser via mcp__browser__ tools: snapshots, clicks, typing, batch list operations, multi-step flows, and tab lifecycle. Use when a task needs real browser state — logged-in sessions, open tabs, visible UI (e.g. "click every chat in my list", "fill this checkout form", "open this in my browser and check it", "continue the tab I handed off"). Prefer web_fetch/web_search for plain reading.
边淬火's general-purpose multi-mentor brainstorming tool. When the user brings a real problem, judgment, or idea they can't currently move forward, first determine which cognitive functions this round needs, then invite multiple high-density mentors whose inputs are isolated within the round and whose knowledge sources differ substantially to expand the thinkable space; the mentors provide viewpoints, concepts, questions, warnings, models, or experience, and the user still forms the final judgment. Applies to "let's brainstorm", "open a round table", "find a few people to talk", "let different people take a look", "淬火 brainstorm". Mentors within a round are blind to each other; later rounds can by default read the full prior mentor texts and user responses. Do not intercept when the user mainly needs learning, diagnosis, logic chaining, decision-making, or execution.
Compatibility entry point for the old name of "Cuihuo Brainstorm". It handles legacy invocations such as /cuihuo-chatroom, "bounce this idea around", "let's talk about an idea" and "idea collision", and passes the user's exact words on to Cuihuo Brainstorm in full. The old semantics of "first step of the content pipeline, the host extracts judgment fragments, automatically entering the logic chain" are deprecated.
Turn vague decisions into a trackable bet card. Four layers: raw information (what happened), decision facts (what you're betting on), hypotheses and inferences (where you might be wrong), and result verification (what actually happened). Two entry points: filing (placing a new bet) and backfilling (an old bet has resolved). A general-purpose tool — usable for decisions in any domain. Trigger: /cuihuo-decision, "help me make a decision", "I've decided", "backfill results"
边淬火's general-purpose problem diagnosis. Helps people see what they can't see, name the sticking points they can't articulate, and get a concrete action prescription. Not limited to business problems — relationships, career, self-perception, ability to act, life direction; applies when the user has already clearly decided to see their own situation, emotions, or behavior patterns. Trigger: /cuihuo-diagnosis, "diagnose me", "help me figure this out", "help me see why I'm like this", or routed in from 淬火思考. When the user only says "I'm stuck" or "I can't figure it out" without a clear type, go to 淬火思考 first.
The third step of the content creation pipeline: polishing a judgment seed into shape — the trio (common-sense stance → reversal → mechanism) plus the nail. The AI performs differential diagnosis, writes prescriptions, presses with follow-up questions to deliver it, and challenges from the opposing side. Not a single word of the judgment's body text is ghostwritten. Trigger: /cuihuo-judgment, "help me polish this judgment", "shape the judgment", "polish the judgment"
Question-shattering tool by 边淬火. Persistent questioning that provokes the asker's unknown unknowns, breaking one big question (or a tangle of questions) into a logical, solvable, dependency-ordered set of new questions. Sub-questions must come from the user's own mouth — AI only asks and compiles, and does not shatter on the user's behalf. One outcome is accepted: after shattering, nothing remains (a pseudo-question is dissolved) — that also counts as success. Triggers: /cuihuo-questioning, "shatter this question", "this question is too big", "I don't know where to start", "help me break this question apart", "keep questioning me"
Systematic study tool by 边淬火. Used to faithfully read through a book, a finite set of texts by one thinker, a school of thought, a body of theory, a book series or a complex knowledge system: it does not require the user to preset a learning goal; it first freezes and fully reads the finite sources, reconstructs the original system's own structure, prepares the complete set of read-through lectures in one go, then accompanies the user in continuous reading; after the first pass it redraws the map and selects nodes for close reading with /cuihuo-learning based only on real questions. Triggers: 「淬火体系」「体系学习」「系统读 X」「系统学习 X」「通读 X」「学一个思想家／学派」「先通览再精读」「继续体系学习」.
Cuihuo Thinking v2, the general thinking entry point and topic hub of the 边淬火 family. It takes unformed input such as "I'm thinking about", "I have a question", "this needs to be thought through", "I'm stuck", "I can't make sense of it" or "I'm thinking about that thing from last time", runs an old-topic check, identity check and language/material/function audit, routes to Investigation, Containment, Dissolution, Cuihuo Diagnosis, Cuihuo Questioning, Cuihuo Verification, Cuihuo Learning, Cuihuo Brainstorm, Cuihuo Logic Chain, Cuihuo Judgment Formation, Cuihuo Decision or GTD, and updates the topic once results come back. It replaces the old "Thinking" skill; do not intercept when the user has already explicitly named a downstream skill.
Concept verification and flipping tool by 边淬火. Helps users see the metaphors, rules, confusions and causal inversions hidden beneath a "taken for granted" concept, and guides them to complete the flip themselves. Two entry points: bring an old concept to verify, or bring an existing flip to deepen. The flip must come from the user's own mouth — AI only asks questions and does not flip on the user's behalf. Triggers: /cuihuo-verification, "verify this concept", "flip this concept", "is this statement right", "help me flip it"
Design and audit customer feedback intake across support tickets, sales calls, app reviews, surveys, NPS, communities, cancellation reasons, feature requests, bug reports, and product analytics. Use when teams need to turn noisy feedback into evidence, prioritization input, roadmap decisions, and customer follow-up without overreacting to anecdotes.
Universal deep research agent team. 13-agent pipeline for rigorous academic research on any topic. 8 modes: full research, quick brief, paper review, lit-review, fact-check, three-way literature scan, Socratic guided research dialogue, and systematic review with optional meta-analysis. Covers research question formulation, Socratic mentoring, methodology design, systematic literature search, source verification, cross-source synthesis, risk of bias assessment, meta-analysis, APA 7.0 report compilation, editorial review, devil's advocate challenges, ethics review, and post-research literature monitoring. Triggers on: research, deep research, literature review, systematic review, meta-analysis, PRISMA, evidence synthesis, fact-check, WHY HOW WHAT papers, 3W literature scan, guide my research, help me think through, 研究, 深度研究, 文獻回顧, 文獻探討, 系統性回顧, 後設分析, 事實查核, 三段式文獻掃描, 引導我的研究, 幫我釐清, 幫我想想, 我不確定要研究什麼, 研究方向, 研究主題.
Experiment executor and monitor for academic research. 2-agent system covering code experiments (ML training, statistical analysis, ETL, simulation) and human studies (surveys, field studies, interviews). 4 modes: run (execute + monitor code), manage (track human studies), validate (statistical interpretation + reproducibility verification), plan (Socratic experiment design). Use when a researcher wants to run or monitor a research experiment, manage or resume a human-subject study, check the statistics or reproducibility of research results, or design an experiment or study, in English or Chinese (e.g., 跑實驗、管理研究、驗證結果、規劃實驗).
Researches Fiverr gigs and sellers via the Crawlora API — search gigs by keyword, pull a gig's full detail (packages, pricing tiers, rating, seller summary), and look up a seller's profile — returning clean JSON. Use when the user wants to find freelance gigs, compare gig pricing/packages, or vet a Fiverr seller.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.