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.
Safely modify specific calibration parameters in the GLM configuration file. Use this skill when you need to update Kw, coef_mix_hyp, wind_factor, lw_factor, or ch while preserving all other settings and respecting published calibration ranges.
Select restaurants across multiple cuisine types (American, Mediterranean, Chinese, Italian) that meet quality standards and budget. Use this when meal planning requires variety across different culinary traditions.
Plan a multi-day road trip across multiple cities without flights, considering driving distances, travel time between destinations, and logical routing to minimize backtracking. Use this when building itineraries that require ground transportation between cities.
Identify major attractions, points of interest, and notable sites across three Ohio cities that appeal to general travelers. Use this when building activities for a multi-city itinerary.
Identify and select pet-friendly lodging options that welcome dogs and meet budget constraints. Use this when travelers have pets and need verified pet-friendly hotel/lodging information.
Distribute a total travel budget across accommodations, meals, transportation, and attractions while tracking spending against real data. Use this when ensuring costs stay within specified limits.
Processes all extracted keyframes by converting them to grayscale, then counts coins, enemies, and turtles in each frame using template matching. Generates a CSV file with frame-by-frame object counts. Use this as the main analysis pipeline after keyframe extraction.
Converts an MP4 video file into individual keyframe PNG images. Use this to decompose a video into analyzable still frames. Extracts one keyframe per scene and stores them in the root directory with sequential naming.
Use INDEX-MATCH with two conditions (series code and year) to retrieve values from a source data table. Apply this when you need to pull specific data points based on multiple criteria from an unstructured data range.
Calculate net exports as a percentage of GDP for multiple countries, then compute descriptive statistics (min, max, median, mean, percentiles) on the percentage values. Use this when analyzing trade performance across multiple entities.
Calculate a weighted average using SUMPRODUCT when weights and values have a one-to-one correspondence. Use this for computing aggregate statistics where different entities have different importance (weights).
Use this skill when you need to apply Anthropic's official brand color palette to any design artifact, poster, or visual. Contains the correct HEX values for all brand color tokens as of 2024.
Use this skill when selecting and applying fonts for Anthropic brand materials. Specifies the official heading and body typefaces with correct fallback chains.
Use this skill when creating a technical exploded-view diagram poster of a hardware device. Covers layout, layer spacing, annotation style, and rendering approach using Python Pillow.
Use this skill when composing a seven-character regulated verse (七言律诗) to ensure correct structural requirements including rhyme scheme, parallelism (对仗), thematic progression, and formatting rules.
Use this skill when composing a seven-character regulated verse (七言律诗) to ensure strict compliance with tonal patterns (平仄). This skill defines the four canonical tonal forms, provides a character-by-character verification method, and enforces a pattern-first composition workflow.
Use this skill when you need to fill out the California Small Claims Court form SC-100. It maps the form's PDF field names to their meanings and expected values.
Use this skill when you need to programmatically fill PDF form fields using Python. Covers inspecting field names and writing values to fillable PDFs.
Use this skill to extract and organize case data from a natural language case description for filling the SC-100 form. Maps case facts to form fields.
Implementing custom distance metrics for DBSCAN in scikit-learn for specialized coordinate-based clustering.
How to perform greedy matching between predicted cluster centroids and ground-truth expert points using closest-pairs-first strategy with a maximum distance threshold.
How to parallelize a grid search over hyperparameter combinations in Python using joblib or multiprocessing for CPU-bound tasks like DBSCAN clustering.
How to identify Pareto-optimal solutions from a set of multi-objective optimization results, specifically maximizing one metric while minimizing another.
How to use offline vulnerability scanning tools like grype, trivy, or osv-scanner to detect vulnerabilities in dependency lock files without network access.
How to perform security audits on Node.js package-lock.json files using npm audit and other offline tools to identify vulnerabilities in third-party dependencies.
How to parse package-lock.json to extract all dependency names and versions for vulnerability lookups, covering both lockfile v1, v2, and v3 formats.
How to generate a properly formatted CSV security audit report from vulnerability scan results, including handling of special characters and proper escaping.
Converting earthquake timestamps from USGS GeoJSON format (epoch milliseconds) to ISO 8601 format for output.
Guidance on selecting appropriate CRS projections for geospatial distance calculations, especially EPSG:4087 for global analyses.
Finding the earthquake furthest from the Pacific plate boundary within the Pacific plate itself, using GeoPandas with EPSG:4087 projection for distance calculations and EPSG:4326 for spatial containment checks.
Core skill for retrieving information from enterprise data files (JSON, JSONL, CSV, Parquet, Markdown, etc.) located at /root/DATA, answering questions from /root/question.txt, and writing structured answers to /root/answer.json. Handles multi-hop reasoning, cross-referencing, and entity resolution across enterprise documents.
How to compare fund holdings between Q2 and Q3 to find top stocks with increased investment by dollar value
Understanding the structure of SEC 13F filing datasets stored in /root/2025-q2 and /root/2025-q3 folders, including available scripts and data files
How to analyze a fund's AUM and stock count using accession_number, with proper filtering of options and deduplication. Covers Q1 (AUM) and Q2 (stock count).
How to fuzzy search for a fund in COVERPAGE.tsv to find its accession_number, using either built-in scripts or manual search
Step-by-step instructions for creating and applying patches to fix the JavaScript RCE vulnerability in Apache Druid 0.20.0. Covers inspecting actual source code patterns, creating correct sed replacements or patch files, and verifying the fix. Use this when you need to patch and build Druid.
Understanding the CVE in Apache Druid 0.20.0 where authenticated attackers can execute arbitrary code via JavaScript payloads. The vulnerability exploits @JacksonInject with empty key "" to override JavaScriptConfig and enable JavaScript execution even when disabled. Use this skill to understand the attack vector and plan the fix.
How to compile PR and issue statistics into a report.json file for a GitHub repository community pulse report, using jq to build the final JSON structure.
How to accurately search for issues (excluding pull requests) in a GitHub repository using the gh CLI, since GitHub's search API treats PRs as a type of issue.
How to accurately search for pull requests in a specific GitHub repository within a date range using the gh CLI, and correctly distinguish merged vs closed-without-merge PRs.
How to set up the correct Python environment for the SimPO project, specifically using Python 3.10 and resolving version conflicts. Use this when setting up the environment to run SimPO code and unit tests.
How to implement the SimPO (Simple Preference Optimization) loss function based on the SimPO paper. Use this when implementing the simpo_loss method in SimPOTrainer.
How to correctly run the SimPO unit test to generate the loss.npz output file with reproducible results. Use when executing the unit test after implementing the loss function.
How to open a .docx Word document template with python-docx, replace placeholder tags like {{PLACEHOLDER}} in all document locations (body, headers, footers, tables including nested tables), handle conditional sections, and save the result.
Detailed guidance on recursively processing nested tables in python-docx documents. Use this when a Word template contains tables inside table cells and you need to reach all paragraphs.
How to extract text from PDF files for content-based classification. Use this when you need to read PDF content to determine what subject/topic a paper belongs to. Covers both text-based and scanned PDFs.
Use this skill to understand the exact Python source code in /root/Tokenizer.py before translating to Scala. This covers every class, method, regex pattern, enum value, default value, and metadata key that must be faithfully reproduced.
Use this skill to read and understand /root/TokenizerSpec.scala to determine the exact expected API signatures, imports, class names, method names, and enum values that the Scala implementation must match.
Use this skill when writing /root/Tokenizer.scala. It provides the concrete implementation strategy, Scala idioms, and compilation workflow for translating the Python Tokenizer to Scala 2.13.
Use this skill when you need to parse meeting schedule request emails from a JSON file, extract meeting duration, date/time constraints, sender information, and message IDs. Handles various natural language patterns for expressing meeting preferences.
Use this skill to extract visual calendar data from a PDF file using PyMuPDF (fitz), including reading text for time labels and appointments, detecting colored rectangular blocks, and measuring their vertical positions to determine start/end times based on a 15-minute-per-row grid.
How to merge GLM simulation output with field observations and compute RMSE metrics for Lake Mendota calibration. Use this when computing overall RMSE, annual deep RMSE, and summer deep RMSE from matched observation-simulation pairs.
How to extract water temperature profiles from GLM NetCDF output files (output.nc). Use this skill when you need to read GLM simulation results, extract depth and temperature arrays, and construct a dataframe of simulated temperatures at specific depths and times.
Use this skill to understand the database file structure and how to search for Ohio cities, restaurants, accommodations, attractions, and distances in the provided dataset files for trip planning.
Use this skill when planning a 7-day road trip itinerary from Minneapolis to three cities in Ohio, covering budget, pet-friendly accommodations, cuisine preferences, and no-flight constraints. Produces a JSON itinerary file.
Use this skill to convert extracted keyframe PNG images to grayscale in-place using Python3 and PIL/Pillow. Only convert keyframe images, NOT template images (coin.png, enemy.png, turtle.png).
Use this skill to count coins, enemies, and turtles in grayscale keyframe images using OpenCV template matching with a provided count_objects.py script. Includes robust output parsing and fallback to direct template matching if the script is unavailable.
Use this skill to generate the final CSV file at /root/counting_results.csv with columns frame_id, coins, enemies, turtles. Frame IDs must be in the format /root/keyframes_%03d.png.
Use this skill to extract keyframes (I-frames) from an MP4 video file using ffmpeg. This produces numbered PNG files in a specified output directory.
Claude Skills — questions
Answers built from the skills we actually parsed.