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...
build-agent
by Largo2z9Custom agent architect. From a fuzzy operator intent, maps the available ecosystem (Context Engine, MCP, existing skills, workspace knowledge), designs the full architecture (1 or N agents), generates SKILL.md files directly executable in the workspace. FR: "construis-moi un agent", "crée un skill", "je veux un agent qui", "build un outil qui", "j'ai besoin d'un agent pour", "crée-moi quelque chose qui". EN: "build an agent", "create a skill", "I want an agent that", "build a tool that".
decompose-ad
by Largo2z9v1.5.0 (v2.64 ontologie sémantique pure · pain_points + objections sub-audience) · reverse-engineer benchmark ad concurrente · si reverse-engineer mentionne pain ou objection détectée dans ad concurrente sur même audience que brand opérée, link vers `audiences/{audience_slug}/pain_points/*.json` ou `audiences/{audience_slug}/objections/*.json` sub-audience canonical SI applicable. Sinon · note descriptive sans canonical ref. Backward compat strict additif · fallback top-level v2.63 + profile sub-fields v1.7 preserved. v1.4.0 (v2.63 ontologie pure · pain_points + objections collections top-level) · reverse-engineer benchmark ad concurrente · si reverse-engineer mentionne pain ou objection détectée dans ad concurrente, link vers `pain_points/*.json` ou `objections/*.json` collections SI applicable (ad sur même audience que brand opérée, signal cross-applicable). Sinon · note descriptive sans canonical ref (ad concurrente = pas notre canonical, observation pure). Backward compat preserved. v1.3.2 (v2.61 doctrine
import-asset
by Largo2z9Orchestrator skill · import + tag + valide un asset brand (logo PNG haute-res, badge cert/claim, mascotte, pattern background, packshot variant) dans le slot canon visual_identity.assets_canonical schema v1.2. Pattern symétrique craft-packshot v1.1 mais SANS génération IA · les brands fournissent leurs assets en général (logo officiel, badges certif obtenus, mascotte design agency). Skill guide operator pour drop fichier local + tag asset_type + validation 5 critères qualité + persist sidecar visual_identity.json avec _validated_by_operator gate. Bridge downstream · compose-creative v1.3+ HR3b layered multi-layer paste consume packshot + logo + badge en couches pour pubs studio photographer avec branding complet pixel-exact. FR: "importe un asset", "ajoute le logo brand", "ajoute un badge cert", "ajoute la mascotte", "ajoute un pattern background", "canonise un asset", "drop asset brand", "upload asset visuel". EN: "import an asset", "add brand logo", "add cert badge", "add mascotte", "add background pattern"
produce-paid-angles
by Largo2z9v1.11.0 (v2.79.5 engagement disclosure NIVEAU 0 paramètres décomposés) · Section pré-runtime ajoutée AVANT Step 0 · expose 6 paramètres décomposés au runtime (audience targeted · pains/JTBD source · formula angles OTRB · couches pain-benefit-chain · hypothèses figées · biais à éviter) avec POURQUOI chacun + close binaire OK ou ajuste. Cross-ref doctrines `docs/system/decomposition-visibility-doctrine.md` v2.79.5+ + `docs/system/engagement-disclosure-doctrine.md` v2.79.5+. Backward compat strict additif (Steps 0-12 runtime preserved · seul l'amont disclosure change). v1.10.0 (v2.64 ontologie sémantique pure · pain_points + objections sub-audience) · Step 1 read encoded data refactor · pain_points lus depuis `audiences/{audience_slug}/pain_points/*.json` (sub-audience NEW v2.64 · owned natif par parent path) · objections lues depuis `audiences/{audience_slug}/objections/*.json` (sub-audience NEW v2.64). Step 11bis back-refs P4/P5 stages canonical refs `audiences/{audience_slug}/objections/{OBJ-NN}.json#response
snapshot-brand
by Largo2z9v1.5.1 (v2.81.1 decomposition visibility NIVEAU LIVE) · NEW section Niveau LIVE thinking aloud obligatoire pendant exécution. Action LOURDE · narratif étendu 2 niveaux abstraction (macro contexte boutique + micro produit many-to-many phrasé spec → mécanisme → bénéfice → pain → audience en prose). Pose pair senior expert · audit temps réel + pédagogie indissociables. Cross-ref `decomposition-visibility-doctrine.md` v2.81.1+ HR-DVD-11 + AP-DVD-11. Backward compat strict additif (cycle runtime préservé). v1.5.0 (v2.78.2 decomposition visibility) · Phase output NEW section Decomposition Visibility matricielle obligatoire APRÈS 5 sections IP · 4 niveaux canon (décomposition produit specs/mécanismes/bénéfices 3 couches · many-to-many pain × audience matrix · stage business filter · méthode pédagogique verbale). Additif strict · existing Phase output preserved. v1.4.0 (v2.68 progressive cartography refactor) · Phase 1 macro confirmation light (3-5 lignes produit+offer+positionning · gate binaire valide/corrige) puis
migrate-workspace
by Largo2z9Migrates an existing brand instance to match a newer template version. Compares instance structure against target template, generates a diff, proposes safe modifications (add fields, rename, deprecate). Never deletes data. FR: "migrate" "upgrade cette brand" "aligne avec le template" "check drift" "mets à jour la structure" "version mismatch". EN: "migrate" "upgrade brand" "align with template" "check drift" "template version mismatch" "sync to latest template".
sync-notion-atlas
by Largo2z9v2.0.1 (v2.67) · territoire-only alignment doctrine territory-doctrine.md NEW. Phase B push refactor · 10 databases territoire strict (Produits, Specs, Mécanismes, Bénéfices, Personae, Pain Points, Angles, Objections, Frictions usage, Roadmap). Full funnel Meta (creatives = production layer) RETIRÉE du push v2.0.1 · creatives push via NEW skill dédié sync-creatives-to-notion v2.68+ futur (production layer séparée · cards/Kanban Notion pour briefs + créas par angle). Phase A pull (Steps 0-6) preserved unchanged backward compat strict additif (pull peut ingérer 11 collections si présent Notion-side legacy). --mode=diff reste deferred Phase C v2.59+. v2.0.0 (v2.66) · Phase B push runtime exec-ready · Steps B1-B7 detailed (canvas + 11 DBs creation + rows population + relations 2-pass + idempotency lookup par phantom_entity_id). Phase A pull (Steps 0-6) preserved unchanged backward compat strict additif. Dual-direction sync operational. --mode=diff reste deferred v2.59+. v1.1.0 (v2.58 coverage extend) · friction.{
recompose-creative
by Largo2z9v1.4.0 (Brique 4 étape C · repath batch + split genome/creative/sidecar + mkdir-claim + variant_axis collapse) · recompose lit le creative SOURCE sous la forme batch `creatives/{batch}/{CRT-NN}/creative.json` (+ genome.json frère pour l'ADN), id source GLOBAL au namespace CRT scanné cross-batch, et PRODUIT une variante = un NOUVEAU CRT-NN dans le batch DU JOUR (split genome.json + creative.json + produced/{slug}.json sidecar + brief.md, dossier réservé par mkdir-claim atomique STEP 0). variant_axis collapse sur l'enum canonique creative.schema (5+null, single). variant_of = le CRT-NN source (namespace CRT, jamais RCV). _schema_version creative/1.4 + genome/1.2. Backward compat · forme cible imposée par le gate. v1.2.2 (v2.61 doctrine consume) · consumes: enrichi avec refs docs/doctrine/ NEW v2.60 (angle-anatomy, hooks-method). Skill peut désormais consume ces doctrines canon copywriting/strategy pour informer production sans dépendre schemas exacts. v1.0.0 (v2.34 ship production loop) : 3e skill P5 visual aux
capture-learning
by Largo2z9Quick append of a single operational learning to learnings.json. Low friction — no full ingest ceremony. FR: "capture ce learning" "note ça" "retiens que" "ajoute dans les learnings" "on a découvert que" "garde en tête". EN: "capture learning" "remember that" "note this" "save this learning" "log this insight".
bird
by Largo2z9Vue d'oiseau sur un projet ou une brand active. Recale l'opérateur instantanément après reprise de session ou perte de fil, en rendant le territoire entier visible (zones acquises, zones bloquées, zones identifiées non engagées, zones non cartographiées). Lit la carte produite par /scope, ne la crée pas. Pair canon avec /scope (scope CRÉE la carte d'un sujet flou · bird LIT la carte). FR: "où on en est", "j'ai perdu le fil", "fais-moi un bird", "vue d'oiseau", "/bird", "/bird all", "/bird --zoom {zone}". EN: "where are we", "I lost the thread", "give me a bird's eye", "/bird", "/bird all", "/bird --zoom {zone}".
import-meta-results
by Largo2z9Pull Meta Insights par ad_id pour creatives produced (CRT-NN) · écrit dans validations[] canon-tools utilisés (formula · framework · archetype · hook · objection · CTA) · alimente decay v2.37 attribution + N≥3 brands threshold auto-promote canon. Ferme la boucle produce → test → learn → promote opérée quotidiennement par skill dédié (vs ad-hoc manuel actuel). Step 0 bridge proactif canon v2.77. FR: "import results Meta", "import perf ads", "feed atlas vivant", "alimente validations canon", "import meta perf", "pull insights canon". EN: "import Meta results", "import ad perf", "feed validations canon", "pull canon results", "import meta insights".
define-brand-voice
by Largo2z9v1.0.0 (v2.80 ship) · Produit brand voice chart canonique brand via méthodologie Nielsen Norman 4D (Funny↔Serious · Formal↔Casual · Respectful↔Irreverent · Enthusiastic↔Matter-of-fact). Output · voice axes scores -5 à +5 par axe avec rationale · do/don't lexique 5-10 entries each sourcés audiences key_expressions + pain_metaphors + solution_metaphors · sample sentences per touchpoint (paid headline short-form · organic caption mid-form · CRM email subject + body · UI microcopy button + tooltip + error). Mute brand.json#/tone_of_voice (extend voice_axes + do_lexicon + dont_lexicon + sample_sentences_per_touchpoint) + crée brand_voice_chart.md brand-side standalone reference card. Step 0 bridge proactif canon v2.77 si territory incomplet (identity + audiences encoded minimum requis). Produce upstream consumer pour validate-brand-voice-consistency post-write. FR · "define brand voice", "tone of voice", "définit la voix brand", "voice chart", "ton de voix", "ancre le tone", "pose le tone of voice {brand}", "c
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.