Claude Skills
The open format is called Agent Skills and works in Claude Code, Codex, Cursor and other agents — most people know it as Claude Skills.
Every Agent Skill we could find on GitHub, deduplicated by content. 79 566 files from 1 758 authors, of which 61 913 are unique — the rest is the same skill repackaged into someone else's repository. For each one: what it weighs in tokens, whether it ships runnable scripts, and which MCP servers it needs.
Formats and saves answers to a JSON file where each entry requires a specific nested dictionary structure containing list-type values and positive numeric token counts.
Extracts and tracks the consumed token usage from an LLM API response to monitor API cost and utilization.
Parses a text file containing key-value pairs (like question IDs and question text) into a Python dictionary.
Guidelines for analyzing 13F filings, finding fund accession numbers, calculating AUM, and counting exact stock holdings by utilizing existing scripts and correctly excluding options.
Techniques for replacing text placeholders, safely deleting paragraphs, and handling custom multi-paragraph conditionals (like IF/END_IF markers) in Microsoft Word documents using python-docx.
Idiomatic translation of Python constructs to Scala 2.13, including classes, interfaces, enums, and naming conventions.
Handling absence of values and errors in Scala using Option, Try, and Either instead of nulls and exceptions.
Implementing the Builder pattern idiomatically in Scala using case classes and the copy method for immutability.
Using regular expressions for text tokenization and parsing in Scala using the scala.util.matching.Regex API.
HTML/CSS/JS patterns for creating tooltips, conditional interactions based on data attributes, and cross-highlighting elements across different views (like a chart and a table).
Use this skill to construct the final itinerary and strictly format the output as a JSON file, enforcing specific key requirements and keywords.
Use this skill to query the local databases for distances, cities, accommodations, restaurants, and attractions to gather real-world data for an itinerary.
How to perform two-dimensional data lookups in Excel using functions like INDEX & MATCH or VLOOKUP & MATCH based on both row and column criteria.
How to calculate statistical summaries (min, max, median, mean, percentiles) and weighted averages using SUMPRODUCT, with specific rules for rounding and avoiding double-scaling of percentages.
Reference guide for Anthropic's official brand colors, typography, and design standards. Use this skill whenever creating branded materials, posters, or visual assets that need to follow Anthropic's corporate identity guidelines. Includes color palette, typography standards, minimalist design principles, and brand application rules.
Guide for generating technical exploded-view posters using Python (PIL/Pillow). Use this skill when creating engineering diagrams, hardware breakdowns, or product teardowns that require precise layer visualization, annotation leader lines, and technical accuracy. Includes methods for drawing components, layering, annotation, and exporting high-quality images.
Guide to modern Mandarin pronunciation-based rhyming for classical Chinese regulated verse. Use this skill when determining rhyme schemes, ensuring proper rhyming sounds in couplets, and verifying phonetic compatibility in classical poetry composition.
Comprehensive guide to seven-character regulated verse (七言律诗) format, structure, and tone patterns. Use this skill whenever composing classical Chinese regulated verse, especially when strict adherence to tonal patterns, line structure, and formal requirements is critical.
Guide to expressing war suffering and peace longing in classical Chinese poetry. Use this skill when composing regulated verse about conflicts, depicting the human cost of war from ordinary perspectives, and yearning for peace through traditional imagery and metaphors.
Fill PDF forms programmatically using pypdf library. Use this skill whenever you need to fill in PDF forms with field data, extract field names from PDFs, or generate filled PDF documents. Essential for automating form completion tasks.
Map case information to California Small Claims Court Form SC-100 fields. Use this skill whenever you need to fill the SC-100 form, understand its structure, extract field names, or map plaintiff/defendant information to the correct form sections.
Compute F1 score and delta (average distance) metrics for Mars cloud cluster evaluation. Use this skill when evaluating DBSCAN clustering results against expert annotations, including handling edge cases like images with no clusters, no expert points, or no matches.
Implement greedy nearest-neighbor matching between cluster centroids and expert annotations. Use this skill when you need to match predicted cluster centers to ground-truth expert points for Mars cloud evaluation, always using standard Euclidean distance with a maximum distance threshold of 100 pixels.
Implement weighted custom distance metric for DBSCAN clustering. Use this skill when working with DBSCAN on Mars cloud data where distances need to be weighted differently across x and y axes using a shape_weight parameter (w). The custom metric is d(a,b) = sqrt((w*Δx)² + ((2-w)*Δy)²), controlling whether x-distances or y-distances are attenuated.
Identify Pareto-optimal solutions from Mars cloud clustering results. Use this skill when you need to find the frontier of solutions that balance multiple conflicting objectives (e.g., maximizing F1 while minimizing delta), eliminating dominated solutions and finding all trade-off points.
Scan npm package-lock.json files for vulnerabilities using Trivy in offline mode. Extracts HIGH and CRITICAL severity vulnerabilities with detailed metadata (CVE, CVSS, fix versions). Use this skill whenever you need to audit npm dependencies for security issues, especially when an offline vulnerability database is available.
Generate structured CSV security audit reports from vulnerability data with proper formatting and schema validation. Use this skill whenever you need to export vulnerability records to CSV format with consistent field ordering, proper escaping, and RFC 4180 compliance.
Process vulnerability scan results from Trivy JSON output, extract HIGH/CRITICAL severity vulnerabilities with complete metadata (CVE, CVSS scores, fix versions, references). Use this skill whenever you need to transform raw vulnerability data into structured format with proper field mapping and CVSS score prioritization.
Master coordinate systems and projections in GeoPandas for accurate spatial calculations. Handle WGS84 to projected coordinate system conversions, select appropriate projections for regions, and perform distance/area calculations correctly. Use this skill when working with latitude/longitude data that needs to be converted to metric distances, or when performing spatial calculations that require specific coordinate systems.
Analyze earthquakes relative to tectonic plate boundaries using GeoPandas. Load earthquake and plate boundary data, identify earthquakes within specific plates, calculate distances to boundaries, and analyze spatial relationships. Use this skill when working with plate tectonics, earthquake location analysis, distance calculations to plate boundaries, or coordinate projection systems in geopandas.
Work with tectonic plate boundary datasets (PB2002 format) to identify plate regions, extract specific plate boundaries, and perform spatial filtering. Load plate boundary and plate polygon data, identify specific plates like the Pacific plate, and extract relevant boundaries. Use this skill when parsing plate tectonics datasets, identifying plate regions from boundary lines, or filtering spatial data by plate membership.
Search and query enterprise JSON data to find specific information like employee IDs, document references, URLs, and insights. Use this skill when you need to locate answers to business questions such as finding authors of reports, team members with specific insights, or shared URLs. Handles multi-step queries, filtering, and deduplication of results.
Load and parse large enterprise JSON data files efficiently. Use this skill when you need to read JSON files from /root/DATA or similar enterprise data directories, especially large files that exceed memory limits. Handles partial reading, searching specific content within JSON structures, and navigating nested JSON hierarchies without loading entire files into memory.
Format enterprise data retrieval results as JSON with token tracking. Use this skill when you need to output query results to JSON files in the format {"q1": {"answer": [...], "tokens": N}, ...}. Handles answer list formatting, token counting, and file output validation.
Analyze SEC 13F fund holdings data given an accession number. Use this skill when you have an accession_number and need to extract fund details including AUM, number of holdings, and detailed position data. Works with Q2 and Q3 2025 filings.
Search for hedge funds in SEC 13F filings by fuzzy name matching. Use this skill whenever you need to find a fund's accession number by name (e.g., "Renaissance Technologies", "Berkshire Hathaway"). Returns accession_number to use with fund analysis.
How to patch the Apache Druid 0.20.0 JavaScript sandbox bypass vulnerability. Use this skill when fixing CVE in Druid where authenticated attackers can execute arbitrary code through malicious JavaScript payloads via the sampler endpoint. The vulnerability allows empty key "" in filter specifications to bypass JavaScript security settings.
How to build Apache Druid efficiently with Maven while skipping unnecessary checks and OOM-prone modules. Use this skill when building patched Druid versions, when rebuilding after security fixes, or when the web-console module causes out-of-memory errors.
How to create, apply, and verify security patches for git repositories. Use this skill when creating patches from code changes, applying patches to source code, or validating that patches correctly address vulnerabilities.
Identify bug reports by checking if any issue label contains the substring "bug" (case-insensitive matching). Use this skill when categorizing GitHub issues as bugs vs. features, filtering defects, or generating quality/reliability metrics.
Query GitHub API for pull requests and issues within a date range, retrieving all metadata needed for metrics collection. Use this skill whenever you need to fetch GitHub PR or issue data with date filtering, especially when gathering activity reports or computing statistics across a time period.
Aggregate GitHub PR and issue data into structured metrics, including contributor analysis, merge statistics, and bug categorization. Use this skill when compiling activity reports, computing open-source velocity metrics, or generating community pulse summaries.
Calculate average time-to-merge from PR creation to merge timestamps, handling edge cases like unmerged and closed PRs. Use this skill whenever computing PR velocity metrics, measuring merge turnaround, or analyzing PR lifecycle duration.
Set up Python environment for PyTorch-based NLP projects with transformers and alignment training. Use this skill when initializing project environments, managing dependencies from environment.yml files, installing required packages, and ensuring CUDA/device compatibility. Essential for reproducible machine learning research requiring specific package versions.
Run PyTorch unit tests, save results to NumPy files, and log environment information for reproducibility. Use this skill when executing test suites for neural network functions, validating loss computations, saving tensor outputs for verification, and creating reproducibility logs with Python/package versions.
Guidelines for implementing the SimPO (Simple Preference Optimization) loss function for NLP model training. Use this skill when modifying trainers or implementing preference-based loss functions in transformer projects.
Generate filled offer letters from a JSON employee data file and a Word template. Use this skill whenever you need to create offer letters by matching data fields to document placeholders, handling conditional relocation packages, and saving the final document.
Replace placeholders in Word documents (.docx files) and handle conditional sections. Use this skill whenever you need to fill in {{PLACEHOLDER}} markers in a .docx template, or when working with conditional blocks like {{IF_CONDITION}}...{{END_IF_CONDITION}}.
Organize large collections of files into categorized folders. Use this skill when you need to move or copy 100+ files into multiple destination folders based on classification data. Handles PDF, DOCX, PPTX and other file types while preserving original filenames and content.
Classify documents by subject matter. Use this skill when you need to determine which category (LLM, Trapped Ion & Quantum Computing, Black Hole, DNA, Music History) a PDF, DOCX, or PPTX file belongs to based on its content. Analyzes titles, abstracts, keywords, and text content to accurately categorize documents.
| How to handle type variance in Scala generics when translating from Python's flexible type system. Use this whenever translating Python generic types (TypeVar with covariant/contravariant bounds) to Scala. Covers covariance (+T), contravariance (-T), and invariance for proper type safety.
| How to convert Python mutable dataclasses and enums to immutable Scala case classes, sealed traits, and sealed objects. Use when translating Python @dataclass, Enum, and mutable collection patterns to idiomatic Scala. Covers immutable-by-default patterns, copy with modifications, and sealed hierarchies.
| temporal types (LocalDate/LocalDateTime), and functional tokenization pipelines. Use when implementing tokenizer classes, handling JSON structures, or building fluent APIs.
| How to convert Python Protocol classes (structural typing) to Scala traits and type classes. Use when translating Python's @runtime_checkable Protocol patterns to Scala's nominal typing with traits. Covers trait-based abstraction, implicit type classes, and structural type emulation.
Claude Skills — questions
Answers built from the skills we actually parsed.