Explore AI Agent Skills & Claude Prompts
Discover open-source agent skills for Claude Code, Codex, ChatGPT, and any tool that uses SKILL.md.
Enter through keywords, occupations, creators, and GitHub sources to see what kinds of skills are emerging across domains.
Use the same catalog through the API
Connect 381,784 public skills to your own search, analytics, or agent workflow with the REST API.
Querying local SQLite index...
infra-advisor
by Soap2GUse when the user describes a goal at the CERN/HEP infrastructure level and wants to know WHICH services to stitch (e.g. "I want distributed analysis on Open Data and to train an ML model"). Returns a 2–4-service stack with steps and pointers to the matching execution skill (`rucio`, `reana-workflows`, `physlite-basics`, …). Bundles a service catalogue (`reference/catalog.yaml`), pre-cooked recipes (`reference/recipes.md`), and digests on GPU access, SWAN+HTCondor scale-out, and columnar frameworks. Does NOT cover physics methodology, single-tool command-level help, or running specific code (use the dedicated skill named by the recommendation). Disambiguator phrase: CERN service stack composer.
fts-rest
by Soap2GUse when the user runs the FTS3 REST CLI (`fts-rest-whoami`, `fts-rest-server-status`, `fts-rest-transfer-submit`, `fts-rest-transfer-list`, `fts-rest-transfer-status`, `fts-rest-transfer-cancel`, `fts-rest-delete-submit`, `fts-rest-delegate`, `fts-rest-ban`) against an FTS3 endpoint such as `https://fts3-pilot.cern.ch:8446` and needs to submit, list, inspect, or cancel point-to-point transfer jobs between SRM / XRootD / HTTPS / S3 storage endpoints, or query the user's FTS identity / server status / delegated proxy. Assumes the FTS endpoint is reachable, a valid X509 proxy (`X509_USER_PROXY`) exists, and `fts-rest-whoami -s <endpoint>` succeeds. Does NOT cover Rucio-driven transfers (Rucio uses FTS3 internally — use `rucio` for rule-based replication), FTS3 docs / configuration / installation (use `cern-docs` with `source="fts"`), or grid job submission (use PanDA / DIRAC). Disambiguator phrase: fts-rest-transfer-submit hyphenated CLI.
rucio
by Soap2GUse when the user runs the Rucio v38+ CLI on lxplus, SWAN, or a CVMFS-staged client (under `/cvmfs/sw.escape.eu/rucio/<version>/`) and needs to query DIDs, RSEs, replicas, metadata, or replication rules for collaboration-internal data. ALWAYS noun-verb (`rucio rse list`, `rucio did show`, `rucio rule add`); NEVER the deprecated flat verbs (`list-rses`, `list-dids`, `add-rule`, `get-metadata`, `rule-info`). Targets non-public datasets — for ATLAS Open Data DSIDs use `atlas-opendata` instead; for grid submission use PanDA, not Rucio directly. Disambiguator phrase: rucio v38 noun-verb.
atlas-opendata
by Soap2GUse when the user has an ATLAS DSID, `physics_short` name, or ATLAS Open Data release tag (`2024r-pp`, `2025r-evgen-13tev`) and needs file URLs, cross-sections, k-factors, filter efficiencies, sumOfWeights, or MC weight metadata. Backed by the `atlasopenmagic` MCP. Operates on ATLAS Open Data only — for collaboration-internal datasets use `rucio`; for non-ATLAS portal records identified by `recid` or DOI use `cern-opendata`. Disambiguator phrase: atlasopenmagic DSID resolver.
cern-opendata
by Soap2GUse when the user has a CERN Open Data `recid`, DOI, or exact record title and needs portal-level metadata, file URIs, container images, or supplementary files across CMS, ATLAS, LHCb, ALICE, or OPERA. Backed by the `cernopendata` MCP at opendata.cern.ch (records, files, glossary). Does NOT cover ATLAS Monte Carlo metadata by DSID or `physics_short` (use `atlas-opendata`), collaboration-internal datasets (use `rucio`), or canonical CERN service / ATLAS-software documentation (use `cern-docs` for SWAN, HTCondor, Cloud, ML@CERN, Athena docs). Disambiguator phrase: opendata.cern.ch recid resolver.
hepdata
by Soap2GUse when the user wants the numerical tables, ROOT files, or YAML data attached to a published HEP measurement, identified by HEPData submission ID (`ins1234567`, `123/v2`), INSPIRE record id, paper DOI, or paper title — typically because they want to re-fit, plot, or run systematics studies on the published measurement. Routes to https://www.hepdata.net/ via WebFetch and the HEPData REST API. Returns table URLs (YAML, ROOT, CSV) suitable for downstream `pyhf`, `pandas`, or `uproot`. Does NOT cover the paper's narrative or the headline measured value (use `read-publication`), PDG canonical constants (use `pdg-lookup`), or Open Data primary datasets (use `cern-opendata` or `atlas-opendata`). Disambiguator phrase: HEPData published-tables retrieval.
read-publication
by Soap2GUse when the user gives you a PDF path, arXiv ID (`2401.12345` or `arXiv:2401.12345`), INSPIRE record id (`recid:1234567` or `inspire-hep:1234567`), DOI (`10.1103/...`), or HEP paper URL and wants extracted content — abstract, measured cross-section / branching ratio / mass, conditions, cuts, fit method. Routes to the right retrieval (`pdftotext` for local PDFs; INSPIRE + arXiv APIs via `WebFetch` for IDs and URLs). Always cites per AGENTS.md rule 5. Does NOT cover PDG canonical particle constants (use `pdg-lookup`), HEPData numerical tables for re-fitting (use `hepdata`), CERN service operator docs (use `cern-docs`), or Open Data dataset records (use `cern-opendata` or `atlas-opendata`). Disambiguator phrase: published-HEP-result extractor.
atlas-notebooks
by Soap2GUse when the user names a specific ATLAS Open Data outreach notebook (e.g. `HZZAnalysis.ipynb`, `Find_the_Z`, `HyyAnalysis.ipynb`) or asks which runtime (Binder / Colab / SWAN / Docker image) to pick for one. Routes to the canonical path under `atlas-outreach-data-tools/notebooks-collection-opendata` and the matching docs section in `13TeV25Doc`. Does NOT cover Standard Model walkthroughs by physics process (use `sm-analyses`), DSID or cross-section lookups (use `atlas-opendata`), or fuzzy infrastructure goals (use `infra-advisor`). Disambiguator phrase: outreach notebook routing.
sm-analyses
by Soap2GUse when the user names a Standard Model process and wants to "rediscover" it with the public ATLAS Open Data — Z→ll, H→ZZ→4l, H→γγ, H→μμ, H→bb, t-tbar, or WZ→3l+ν. Pairs the process to its docs section in `13TeV25Doc/StandardModel` and its notebook in `notebooks-collection-opendata/13-TeV-examples/uproot_python/`. Does NOT cover BSM searches, generic notebook indexing by filename (use `atlas-notebooks`), or analysis-code authoring (use `physlite-basics` or the `analysis` agent). Disambiguator phrase: SM walkthrough by process.
cern-docs
by Soap2GUse when the user asks how a CERN service or ATLAS / FTS framework operates and the answer should come from canonical operator documentation — HTCondor / lxbatch submission, SWAN session settings and HTCondor pool, CERN Cloud (OpenStack) flavors and quotas, ML@CERN training and serving, ATLAS Athena / ASG / Tier-0 / databases, ATLAS computing and grid production, FTS3 (File Transfer Service) configuration / installation / REST API / messaging / monitoring. Backed by the `cerndocs` MCP (`search_docs` for BM25 search, `fetch_doc` for one page) over 8 indexed sources: atlas-sft, atlas-computing, atlas-databases, batch, cloud, ml, swan, fts. Does NOT cover the `fts-rest-*` CLI itself (use `fts-rest`), Open Data dataset / recid / DOI lookup (use `cern-opendata` or `atlas-opendata`), live job-state inspection (use `reana` or shell tools), or multi-service infra recommendations (use `infra-advisor`). Disambiguator phrase: CERN docs BM25 multi-source.
pdg-lookup
by Soap2GUse when the user asks for a canonical Particle Data Group value — particle mass, lifetime, branching ratio, decay width, magnetic moment, mixing parameter, charge, spin, or any quoted "PDG average" constant. Backed by https://pdg.lbl.gov/ via WebFetch (HTML pdgLive pages and the PDG REST API where available). Always cites the PDG record URL and edition year. Does NOT cover ATLAS Monte Carlo metadata (use `atlas-opendata`), measured values from a specific paper (use `read-publication`), HEPData tabulated measurements (use `hepdata`), or conceptual physics explanations of why a particle has the value it does (use the `tutor` agent). Disambiguator phrase: PDG particle data group lookup.
physlite-basics
by Soap2GUse when the user needs to OPEN an ATLAS DAOD_PHYSLITE file with `uproot` / `awkward`, list its branches, reconstruct calibrated objects (electrons, muons, photons, taus, small-R jets, large-R jets, MET, trigger decisions), or apply MC normalisation. Assumes a public HTTPS or XRootD URI such as `root://eospublic.cern.ch/...` from `atlas_get_urls` or `cod_list_files`. Does NOT cover the older flat 13 TeV reduced ntuples (use the matching tutorial via `atlas-notebooks`), CMS NanoAOD or MiniAOD, or sample discovery (use `atlas-opendata`). Disambiguator phrase: PHYSLITE branch reader.
Browse Agent Skills by Occupation
23 major groups · 867 SOC occupations
Browse by Category
Explore agent skills organized by their primary use case
Explore the agent skills ecosystem by occupation and creator
SkillMD is not just a keyword search box. It is an open map that organizes public skills by occupation, creator, and repository, helping you see which workflows, judgment criteria, and domain habits people are writing for AI agents.
Then follow creators and GitHub repositories back to the source: compare the skills a team maintains, whether the repo is active, and how the README frames the work before you open, install, or reuse anything.
Use it three ways: learn an unfamiliar field by occupation, study how creators organize skills, then use source context to decide what is worth opening or reusing.
01 Map a field
Browse 23 occupation groups and 867 SOC roles to learn what skills exist in adjacent domains and how they break down real work.
02 Follow creators
Use creator and repository pages to inspect maintained skill collections, recent updates, and source context before trusting a result.
03 Search with sources
Search 1.7M+ collected skills, then use occupation tags, creators, and GitHub source context to decide what is worth opening.
Start with the occupation map, then follow creators and repositories back to real code. SkillMD helps explain why a skill is worth opening, not only what it is named.
Standardizing Agent Capabilities with SKILL.md and Model Context Protocol (MCP)
In the rapidly evolving landscape of artificial intelligence, LLM agents (Large Language Model agents) have transitioned from simple text predictors to autonomous problem solvers. To orchestrate complex, multi-step agentic workflows, developers require a standardized format to specify agent capabilities, prompt instructions, system rules, and database bindings. This is where SKILL.md and the Model Context Protocol (MCP) have emerged as standard developer paradigms. SkillMD serves as the central directory for indexing, exploring, and sharing these critical agent configurations.
Our open-source registry currently tracks over 1.7 million collected SKILL.md configurations and system prompts. By compiling agent configurations from active developers on GitHub, we bridge the gap between prompt engineering research and production execution. Whether you are building agents with Anthropic's Claude Code, OpenAI's GPT-4, Google's Gemini, or local models using Ollama and LlamaIndex, standardized skill definitions ensure your agents behave predictably across different runtime environments.
What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open-source standard designed to connect LLMs to data sources, developer tools, and external environments. MCP establishes a bidirectional communication channel between client applications (like Cursor, Claude Desktop, or custom agent systems) and servers hosting data or capabilities. Standardizing instructions via SKILL.md enables LLMs to query databases, read local files, execute terminal commands, and integrate third-party APIs. SkillMD allows you to find ready-to-run MCP servers and prompt instructions for various occupations and technical tasks.
The Structure of a Professional SKILL.md File
A valid SKILL.md configuration is designed to be easily read by humans and parsed by LLMs. It contains precise system instructions, trigger conditions, required parameters, and execution examples. Below is the typical architectural blueprint of a professional agent skill:
- Metadata & Core Scope: Declares the name of the skill, author details, target models, and a description of the capability.
- Triggers & Intent Detection: Details semantic triggers that help the agent decide when to invoke this skill.
- System Prompts: Explicit system-level instructions that direct the agent's behavior, personality, safety guardrails, and formatting preferences.
- Capabilities & Tools: Lists the files, databases, or APIs the agent must access to complete the tasks.
- Few-Shot Examples: Demonstrates real inputs and outputs, helping the model generalize behavior through in-context learning.
Optimizing Agent Workflows for Modern LLMs
Writing effective agent skills requires deep knowledge of prompt engineering. With the release of advanced reasoning models like Claude 3.5 Sonnet, ChatGPT o1, and DeepSeek-V3, prompt templates must focus on structured thinking. Developers are encouraged to use XML tags (e.g., <thought>, <context>, and <rules>) to isolate execution boundaries. Standardized prompts prevent agents from suffering from context drift, ensuring that long-running tasks remain aligned with the initial system parameters.
Exploring by SOC Occupations and Creator Profiles
What makes SkillMD unique is its taxonomy. Instead of simple text search, we parse and organize files according to the Standard Occupational Classification (SOC) system. This means you can discover skills written for Computer and Mathematical roles, Business and Financial operations, Legal, Design, and and Educational Instruction fields. By tracking creator profiles, developers can study how different teams organize their custom instructions, compare version updates, and fork public configs for specialized enterprise use cases.
SkillMD operates as a high-performance index running on a fast Go backend and a highly responsive Astro SSR frontend. All search queries execute in milliseconds, featuring smart debouncing to prevent multiple API requests while keeping user data secure. Join our community of developers to standardize your AI agent instructions and optimize your LLM prompting workflows today.
Frequently Asked Questions
A practical guide to agent skills: what they are, how to inspect them, and how SkillMD helps you explore the ecosystem.