Skills
96,770 packaged capabilities and playbooks for agents. Install any of them with one command.
Data & databases
Run comprehensive quality checks on tabular data (CSV/Excel/TSV/JSON), detecting missing values, duplicates, outliers, format issues, and type inconsistencies to produce an overall score, grade, and actionable suggestions. Triggered when users ask to check data quality, find missing or duplicate values, detect outliers, validate formats, profile data, or clean data.
Extract clean markdown content from web pages using Defuddle CLI, removing clutter and navigation to save tokens. This skill MUST be used instead of WebFetch when the user provides a URL to read or analyze, for online documentation, articles, blog posts, or any standard web page. MUST NOT be used for content already served as Markdown; identify it by the response content type and actual body rather than the URL suffix, and read it directly with WebFetch or curl.
Upload local files to a Dify pipeline knowledge base through the published pipeline datasource flow and optionally apply existing metadata fields. Use when the target dataset is a published rag_pipeline/local_file knowledge base or ordinary Dify document create APIs return completed with 0 chunks.
Handles Entity Framework Core database access, migrations, and repository patterns for PostgreSQL. Use when: Creating DbContext classes, writing migrations, implementing repositories, configuring entity relationships, or optimizing database queries.
Design statistically rigorous A/B tests and interpret experiment results. Use when asked to design an experiment, run an A/B test, calculate sample size, interpret test results, or assess whether an experiment was successful. Produces a complete experiment design with hypothesis, sample size, run time, success criteria, and risk flags — or a results interpretation with ship/iterate/kill recommendation.
Plan, configure, inspect, restart, and analyze bounded FluidSim computational-fluid-dynamics simulations with explicit numerical-validity and HPC safety checks. Use for FluidSim solver selection, parameter review, FFT/MPI setup, output diagnostics, or restart compatibility.
Use when the user wants to connect a GA4 (Google Analytics 4) property, set up GA4 authentication/login, or fix GA4 access. Marketer-friendly OAuth login (no gcloud) via a client-provided desktop OAuth client, then property selection. Produces .ga4/token.json and .ga4/config.json for the ga-use skill.
Use when the user wants to analyze or query GA4 (Google Analytics 4) data — event counts, users, sessions, page views, top-N rankings, filtering by dimension/metric, and period-over-period comparisons via the Data API runReport. Consumes .ga4/config.json + token.json from the ga-auth skill and optionally .ga4/taxonomy.md from ga-taxonomy.
Use when the user mentions Google Analytics / GA4 but the request is vague ("take a look at our GA", "how's our data?", "GA 좀 봐줘", "우리 데이터 어때?"), is getting started ("where do I start?", "뭐부터 해야 해?", right after installing), asks what these skills can do, or makes a multi-step request spanning connection, event taxonomy, and analysis (e.g. "parse our taxonomy sheet and analyze last month", "정의서 올리고 지난달 분석해줘"). Diagnoses the .ga4/ journey state and routes to the ga-auth, ga-taxonomy, or ga-use skill in the right order. Clear single-purpose requests (a specific analysis, parsing a taxonomy file, logging in) are handled directly by those skills — do not use this one.
Provides geospatial and Earth observation workflows using GeoPandas, Rasterio, GDAL, Xarray, Shapely, Laspy, PDAL, Google Earth Engine, and STAC/Planetary Computer, covering Sentinel, Landsat, MODIS, SAR and hyperspectral imagery, spectral indices, terrain and network analysis, point clouds, COGs, CRS handling, and spatial ML, with code in Python, R, Julia, JavaScript, C++, Java, Go, and Rust. Use when computing NDVI or other indices from satellite imagery, when running vector overlays, reprojection, or spatial statistics on shapefiles, GeoJSON, or GeoPackage data, when searching STAC catalogs and reading cloud-optimized GeoTIFFs, when classifying land cover or training ML models on Earth observation data, or when doing terrain, hydrology, or point cloud analysis. Not for general non-spatial data analysis or tabular ML.
Takes a competitor's public GitHub repo URL, fetches their open issues via the GitHub REST API, filters noise locally, clusters issues into 6 demand categories, computes a demand score per issue and per cluster, and outputs a ranked demand gap report with a GTM messaging brief. Use when asked to scan a competitor's GitHub issues, find what their users are begging for, turn GitHub complaints into product positioning, identify competitor feature gaps, or generate messaging from real user demand. Trigger when a user says "scan competitor issues", "what are users asking for on X repo", "find demand gaps in Y", "turn GitHub issues into messaging", or "what should I build based on competitor complaints".
This skill helps users automatically scrape business data from Google Maps using the BrowserAct Google Maps API. Agent should proactively trigger this skill for needs like finding restaurants in a specific city, extracting contact info of dental clinics, researching local competitors, collecting addresses of coffee shops, generating lead lists for specific industries, monitoring business ratings and reviews, getting opening hours of local services, finding specialized stores (e.g., Turkish-style restaurants), analyzing business categories in a region, extracting website links from local businesses, gathering phone numbers for sales outreach, mapping out service providers in a specific country.
Work with a Grafana instance — search and read dashboards, run datasource queries (Prometheus, Loki, PostgreSQL, etc.), inspect alert rules and silences, post annotations, and manage folders. Use this whenever the user mentions a Grafana dashboard, panel, or alert; pastes a Grafana URL; asks "what does this dashboard show", "query this metric in Grafana", "is this alert firing", "silence this alert", or wants to create/export a dashboard — even if they don't say "API". Always start from this skill when interacting with this service — its bundled scripts and recipes are the fastest path.
Expertise in architecting, implementing, reviewing, and debugging hierarchical matching systems. Use when working with: (1) Two-sided matching (Gale-Shapley, hospital-resident, student-school), (2) Assignment/optimization problems (Hungarian algorithm, bipartite matching), (3) Multi-level hierarchy matching (org charts, taxonomies, nested categories), (4) Entity resolution and record linkage across hierarchies. Triggers: debugging match quality issues, reviewing matching algorithms, translating business requirements into constraints, validating match correctness, architecting new matching systems, fixing unstable matches, resolving constraint violations, diagnosing preference misalignment.
Text network analysis and knowledge graphs via the InfraNodus MCP server: analyze text structure, find content gaps, generate research questions and ideas, compare texts, optimize content for SEO/GEO, analyze Google search results, GraphRAG retrieval, structured memory. Accepts text, URLs (YouTube transcription included), and saved graphs. Also builds knowledge graphs of code repos and Obsidian vaults: invoked in a project folder ("graph this repo", "analyze this vault", "/infranodus") it mines docs, PDF text, docstrings, WHY/NOTE comments, and commit/PR/issue history into saved graphs with a report, and keeps them current by replacing the statements of changed files in place. When infranodus/manifest.json exists in the project root, answer questions about themes, decisions, rationale, or gaps by querying the existing graphs FIRST, before reading files. For an LLM-authored knowledge base built from sources (wiki pages, curated ontologies), prefer the llm-wiki skill.
Design effective KPI dashboards with metrics selection, visualization best practices, and real-time monitoring patterns. Use this skill when building an executive SaaS metrics dashboard tracking MRR, churn, and LTV/CAC ratios; designing an operations center with live service health and request throughput; creating a cohort retention analysis view for a product team; or debugging a dashboard where metrics contradict each other due to inconsistent calculation methodology.
Light research mainline, step 9 — figures: make each figure serve an argument (which claim it supports and what would be missing if removed); set publication-grade specifications (column width, font size, colorblind safety, error bars, n, significance); ensure visual honesty (no secretly truncated y-axes, no false correlations from dual y-axes, no jet-rainbow); multimodally “take a real look” after rendering; research data figures must be generated programmatically, never with AI image generation. When to use: after results analysis, to draw paper figures; plan a figure set (figure list, budget, redundancy control); choose chart types; create publication-grade figures (column width, font size, dpi, colorblind-safe palette); address concerns that a figure is misleading or presents a nonsignificant result as the main figure; inspect rendered output for overlapping labels or legends obscuring data; create framework or schematic diagrams (programmatic, not AI-generated). Triggers: draw / figures / figure / illustration / render / paper figure / main figure / caption / error bar / colorblind / viridis / truncated y-axis / dual y-axis / heatmap / framework diagram / panel / display item / column width / publication-grade / chart recommendation / review render / plot / chart / matplotlib / seaborn. Core rules: **visual dishonesty (truncated y-axis / false correlation from dual y-axes) = critical gate** (spec §4.2, STAGE_GATES[9]=[visual_honesty]); missing error bars / jet-rainbow / over-budget figure set / redundancy = warn; **presenting a nonsignificant result as the main figure / mismatch between figure and text = critical → initiate 9→7 rework**; **research data figures must be generated programmatically (matplotlib/seaborn/R), never with AI image generation** (permanent rule); static lint only flags suspicious forms—whether something is truly misleading requires final judgment based on the caption and a person/reviewer. This skill initiates **9→7 rework (nonsignificant result as main figure / figure-text mismatch → result-analysis)**.
Use this skill to classify tabular data end-to-end in MATLAB — load a dataset, prepare and clean it, select promising classifiers, train them, and compare accuracies with cross-validation, holdout, or hyperparameter optimization plus statistical tests. TRIGGER when: user asks to classify tabular data, pick classifiers for a dataset, compare classifier accuracy, run cross-validation or a holdout evaluation, or find the best model with statistical uncertainty. DO NOT TRIGGER when: user has non-tabular inputs (images, sequences, time series), wants a regression model, is training a specific neural network architecture (use matlab-train-network), or wants cost-sensitive learning or an arbitrary class-prior vector (this skill only supports the built-in uniform-prior toggle for imbalanced data).
Connects MATLAB to Databricks using JDBC drivers via Database Toolbox. Use when creating a JDBC connection to a Databricks cluster or SQL Warehouse, configuring Databricks authentication (PAT, OauthU2M, OauthM2M), selecting between Simba and OSS JDBC drivers, using databricks.JDBCConnection, StandaloneJDBCConnection, databricks.SQLWarehouse.connect(), or optimizing Databricks write performance.
Use when importing or exporting vehicle data from/to log files (MDF/MF4/DAT, BLF, ASC/TXT), decoding CAN/CAN FD/LIN messages to signals via DBC, ARXML, or LDF databases, writing timetable data to MDF or BLF files, or calling blfread, blfinfo, blfwrite, mdfRead, mdfWrite, mdfCreate, mdfInfo, canSignalImport, canMessageImport, canMessageTimetable, canFDMessageTimetable, canSignalTimetable, or linMessageTimetable.
Computes the weekly category state for a Yahoo H2H Categories matchup across all 10 scoring categories (R, HR, RBI, SB, OBP, K, ERA, WHIP, QS, SV). Pulls current totals from Yahoo, builds rest-of-week per-cat projections from roster + schedule, then DELEGATES matchup/per-cat win-probability math to `matchup-win-probability-sim`. Consumes the sim's `per_cat_win_probability` and `matchup_win_probability` to derive cat_position, cat_pressure, cat_reachability, and cat_punt_score, and emits a "push 6, punt N" plan that drives waiver, streaming, and lineup decisions. Use when user asks about "category state", "where am I winning", "should I punt", "matchup score", "cat pressure", weekly category planning, or which cats to push vs. concede.
Deploy and consume the Multimodal Embedding Serving microservice — bring it up with setup.sh + docker compose (from a repo clone, or by fetching those same files from GitHub when no clone exists) using the prebuilt intel/multimodal-embedding-serving image, embed text/images/videos over REST on port 9777, choose among 19 models (CLIP/SigLIP/MobileCLIP/CN-CLIP/Blip2/ QwenText), or integrate in-process via the Python SDK wheel. Use when an app needs embeddings for similarity search or retrieval. Not for modifying the service's source — that is multimodal-embedding-serving-dev.
Monitor, query, and manage New Relic observability data via the newrelic CLI. Covers NRQL queries, AI Monitoring telemetry, APM performance triage, deployment markers, read-only alert and incident inspection, infrastructure monitoring, agent diagnostics, and identity/access administration. Use when user asks about LLM token usage or AI application telemetry, application performance, error rates, slow transactions, deployment tracking, incident inspection, New Relic setup, users, groups, roles, permissions, or access grants.
Explore a website once, find the stable data path, and deploy it as a reusable, parameterized Notte Function (a callable HTTP endpoint that can be scheduled and run at scale). Use when a user wants to forge, build, generate, or "bake" a reusable scraper or browser automation for a site, turn a working scrape into an API/endpoint/job, or says things like "I'll run N keywords later", "make this reusable", "pull all listings", "automate this site every day", or "build a function that extracts X from Y". Pairs with notte-functions-doctor, which repairs a forged Function when the site changes.
Procedure for extracting structured data from PDFs with opendataloader-pdf (ODL) correctly: read the installed tool's own help to discover its current options, build the minimal command for the user's goal, verify the extraction actually succeeded, and diagnose silent failures the tool does not report. Use when the user is using, evaluating, or considering opendataloader-pdf/ODL to extract, parse, or convert PDF content to text, markdown, JSON, or HTML — including scanned-PDF OCR, tables, bounding boxes, or a RAG pipeline over PDFs. Do NOT use for PDF merge/split/rotate, Office-format conversion, form filling, or PDF/UA accessibility-compliance tagging.
Formulate and implement operations research optimization models—LP, MIP/QP, constraint programming, network flows, assignment, VRP, scheduling, resource allocation, inventory/production planning; heuristics and metaheuristics; sensitivity and infeasibility diagnosis; solver integration (OR-Tools, Gurobi, CPLEX, HiGHS, PuLP, Pyomo); production OR APIs (timeouts, warm starts). Use for operations research, OR engineer, optimization model, linear programming, mixed integer programming, MIP, VRP, vehicle routing, scheduling optimization, OR-Tools, Gurobi, constraint programming, resource allocation optimization, infeasible model, metaheuristic—not ML prediction (data-scientist), SCM strategy without optimization math (supply-chain-manager), WMS features (wms-developer), simulation platforms (simulation-software-engineer), generic backend (senior-software-engineer), dbt/warehouse (analytics-data-engineer).
PostgreSQL operations, optimization, and administration. Use for: schema design, index selection, query tuning with EXPLAIN ANALYZE, postgresql.conf configuration, backup and restore (pg_dump, pg_basebackup, WAL, PITR), vacuum and autovacuum tuning, connection pooling (pgBouncer, pgPool), replication (streaming, logical), partitioning, monitoring (pg_stat_statements, pg_stat_activity), JSONB operations, full-text search (tsvector, tsquery), row-level security (RLS), extensions (PostGIS, pg_trgm, timescaledb), GiST/GIN/BRIN indexes, materialized views, foreign data wrappers, LISTEN/NOTIFY.
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.