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...
extract-theme
by FirzusExtracts the visual theme (colors, typography, radius, spacing, shadows) of a public website provided by the user and produces a shadcn/ui + Tailwind CSS v4 compatible token block, ready to paste into the project's `globals.css` (or `app.css`) inside `@theme` and `.dark`. Uses the chrome-devtools MCP server to drive a real Chromium instance (no Playwright). Use when the user asks to "extract the theme of <url>", "reverse-engineer the design tokens of <site>", "copy the look of <site> into shadcn", "build a tailwind v4 theme from <url>", or mentions extract-theme / theme extraction / design tokens from a URL.
vite-plus-best-practices
by FirzusProvides best practices for Vite+ (vp), the unified web toolchain combining Vite, Vitest, Oxlint, Oxfmt, Rolldown, tsdown, and Vite Task. Covers the vp command surface (dev, build, check, test, run, pack, install, env), unified vite.config.ts blocks, monorepo overrides, task caching, commit hooks, library packaging, and migrating existing Vite/Vitest/ESLint/Prettier projects. Use when the user mentions Vite+, vite-plus, the `vp` or `vpx` CLI, Oxlint/Oxfmt in a Vite context, tsdown, Vite Task, or asks to configure, migrate, scaffold, or upgrade a Vite+ project.
traversal-system
by FirzusArchitecture blueprint for the traversal layer above the character controller in open-world and action games: world traversability data (climbable-by-default markup, runtime surface probing, traversal volumes, anchors, readability and telegraphing), composable traversal verbs (systemic climbing, glide, swim/dive, grapple, mounts, regional verbs) AND the parkour/momentum tradition (wall-run, free-running, web-swinging, the momentum-vs-stamina opposite economies, the automation-vs-expression tension), the technical implementation (mantle/vault detection trace cascades, motion warping, procedural IK, surface probing at scale, wall-run and grapple physics, custom movement modes and networking), mounts and vehicles (bonding, flying-mount trivialization, build-your-own vehicles, multi-crew sailing, traversal-as-loop), stamina as the open-world governor, and the design valves against traversal trivializing content. References: BotW/TotK and Genshin (the systemic school), with Mirror's Edge, Titanfall 2, Spider-Man,
character-controller
by FirzusArchitecture blueprint for character controllers in production games, third-person AND first-person: kinematic collide-and-slide movement solver, hierarchical movement state machine (ground/air/climb/swim/glide), ground handling (slopes, steps, snapping, moving platforms), jump parametrization and game-feel numbers (coyote time, input buffering, gravity multipliers), stamina economy, modular movement abilities, the animation interface, network-ready structure; the FPS/momentum movement model (the Quake/Source friction+acceleration physics, bunnyhopping/ strafe-jumping/surf math, movement-shooter momentum — wall-running, slide- hopping, the lurch, Doom/Tribes/Mirror's Edge, first-person camera concerns — eye anchor, head bob, FOV-on-sprint, slide/crouch/mantle, vehicle and mounted controllers, fast-movement netcode and anti-cheat); and the locomotion-animation/feel/accessibility layer (blend trees and foot IK, motion matching, procedural and active-ragdoll animation, game feel and juice, movement accessibility
minimap-worldmap
by FirzusArchitecture blueprint for minimap and full-screen world map systems in open-world games AND across genres: map content pipeline (automated orthographic bake with stylization pass, tiled zoom pyramid), the single world-to-map transform asset, a shared marker registry serving both surfaces (categories, zoom-LOD tiers, clustering, label collision, pooling), region-based fog of war with reveal policies, multi-layer/ underground maps, pan/zoom/pin interactions, fast-travel integration, breadcrumb trails (Hero's Path), the cartography/GIS rendering tech (slippy-map quadtree tiling, raster vs vector tiles, SDF labels and icons, label-placement collision, hillshade/hypsometric relief, the painted-map shader, large-world double-precision transform, BC7/ASTC tile compression), and the cross-genre UX (RTS minimap-as-command- surface, MOBA minimap-as-macro-game with ping wheels, FPS UAV/radar info-warfare, the anti-minimap/diegetic movement — Far Cry 2 GPS, Ghost of Tsushima guiding wind, Elden Ring minimal markers, Sub
stylized-rendering
by FirzusArchitecture blueprint for stylized / non-photoreal (NPR) cel-shaded game rendering in the anime lineage of Genshin Impact, Honkai: Star Rail, Guilty Gear Xrd, and Zelda BotW. Covers the lighting model (half-lambert + ramp/step lighting, the ILM/LightMap channel-packed control map, tinted shadows, fwidth terminator AA, light-direction overrides), the line kit (inverted-hull outlines with smoothed normals, post-process depth/normal edges, Xrd UV-beam inner lines), character shading (SDF face shadow maps driven by light angle, eyes-through-hair, anisotropic Kajiya-Kay hair, MatCap), materials and post-process (stepped specular, light-masked rim, environment-softer-than- hero, neutral tonemap vs ACES, AA/bloom for flat palettes), the art-direction calibration dial, and Unity 6 (URP/HDRP) + UE5 mappings. Use when building a toon/cel/anime/stylized look, authoring NPR shaders, outlines, ramp lighting, face SDF shadows, or when a stylized character "reads flat", "looks uncanny", shadows look noisy/blotchy, or outli
coop-session
by FirzusArchitecture blueprint for drop-in/drop-out co-op (2-4 players) in open-world action RPGs, modeled on Genshin Impact's lobby system, plus the wider netcode and co-op-design landscape: the session lifecycle (unlock gates, world-level join rules, join/leave/kick flows, the host's-world-as-server-instance model, late-join state snapshots), the authority spectrum (server-hard economy, client-trusted ephemera — the datamined Genshin hybrid), replication at architecture level (prediction, reconciliation, interpolation, relevance), the co-op content rules matrix (host-only / instanced / shared as data — structural anti-grief), enemy scaling, topology trade-offs (dedicated instances vs listen-server vs P2P); the netcode-model landscape (deterministic lockstep for RTS, rollback for fighting games, client-server lag compensation / favor-the- shooter for FPS, MMO-scale interest management / sharding / single-shard time dilation / server meshing, matchmaking and SBMM — Elo/Glicko/ TrueSkill, relay vs dedicated, NAT trave
inventory-equipment
by FirzusArchitecture blueprint for inventory and equipment systems across gacha, ARPG/looter, and online/MMO games: the instance-vs-count data model (stable GUIDs, polymorphic tabs, caps as policy), gear generation (the datamined Genshin 4-draw RNG pipeline AND ARPG affix systems — prefix/suffix slots, ilvl-gated tiers, weighted pools — plus the deterministic↔random crafting spectrum: PoE currency orbs, D4 tempering/masterworking, Last Epoch forging potential, runewords), unified enhancement (per-type tables, fodder recycling, lock-as-invariant), inventory UI (grid/list/Tetris layouts, loot-filter DSLs, search syntax, stable sorts, compare tooltips, loadouts and transmog), and networked persistence (server-authoritative ownership, duplication prevention via 2-phase-commit trades + escrow + idempotency + append-only ledgers, real dupe postmortems). References: Genshin (Grasscutter schema), Diablo II/III/IV, Path of Exile, Last Epoch, Destiny 2, WoW, EVE, with Unity 6 and UE5 (Lyra) mappings. Use when designing or buil
dialogue-system
by FirzusArchitecture blueprint for dialogue systems across open-world, RPG, and narrative games: the dialogue graph data model (flow/text/conditions as three separate stores, branching-structure patterns, storylet/quality-based and salience-based narrative), the runtime session (input/camera/HUD scope, in-world staging as data, voice-end advancing, interruption policies), barks and the Valve fact-matching model, narrative-design mechanics (skill-check and dice-roll dialogue, dialogue wheels vs full-text, reactivity budgets, delayed consequence, timed/real-time conversation, internal-voice systems), presentation (typewriter, auto/skip, backlog, deep subtitle accessibility), and the authoring/VO/localization pipeline (writer tools — Ink/Yarn/articy, line IDs, lip-sync, RTL/CJK, text expansion, AI/LLM-driven NPCs and the SAG-AFTRA constraints). References: BotW/TotK and Genshin (datamined), Valve GDC 2012, Disco Elysium, Baldur's Gate 3, Mass Effect, Witcher 3, Oxenfree, with Unity 6 / UE5 mappings. Use when designing o
progression-economy
by FirzusSystems architecture blueprint for progression and economy in live-service and single-player games: stat curve tables (base × curve[level], shared curves as data) and the XP-curve families (Pokémon experience groups, RuneScape's geometric curve, D&D thresholds, prestige/paragon infinite progression), skill trees and soft caps / diminishing returns, the stat aggregation contract, multi-currency wallets and the live-ops economy discipline (faucets vs sinks, inflation control, the in-house economist, player-driven markets, wealth concentration), energy gating with timestamp-based regen, battle-pass and season models across the industry (net-positive currency earn-back, FOMO-expiry vs never-expires, retroactive entitlements, idempotent rollover), and the server-authoritative transaction model (validate-commit-notify, idempotency keys, int64 currencies). References: Genshin (datamined + Grasscutter), EVE Online (the economist's craft), WoW, RuneScape, Diablo, Fortnite, Helldivers 2, with Unity 6 / UE5 mappings. Us
save-persistence
by FirzusArchitecture blueprint for game save systems across single-player and online games: the versioned store decoupled from runtime objects (deltas keyed by stable IDs, the world-state store), the save models compared (checkpoint, save-anywhere, continuous souls-style, server-authoritative), the serialization and storage engineering (format comparison — JSON/MessagePack/MemoryPack/ Protobuf/FlatBuffers, schema evolution and the never-reuse-a-field-ID rule, compression, atomic writes and the fsync/rename correctness, stable references, async save performance), networked/MMO persistence at scale (the RAM-of-truth write-back pattern, SQL vs NoSQL backends, sharding, idempotency and ledgers, the rollback playbook, live-service player-data platforms), save UX and cross-progression (slots/autosave, cross-save vs cross-progression, New Game Plus, save-scumming counters, platform certification, cloud quotas), and the tamper-protection ladder. References: Skyrim changeforms, BotW revival flags, souls journaling, Genshin/EV
world-time-weather
by FirzusArchitecture blueprint for time-of-day, weather, seasons, and climate in open-world games: the central game clock service (time scale, dual day divisions, pause rules, dual server/diegetic clocks, calendars and seasonal cycles), weather as data not simulation (climate profiles, pre-rolled regional schedules, the override stack) AND simulation-driven weather (volumetric clouds, atmospheric scattering, dynamic storm systems), systemic weather consumed by traversal/combat/AI/audio (the BotW chemistry-engine model) plus survival temperature and biome/climate systems, the event scheduler (blood-moon-style resets, respawn policies, time-skip catch-up), the time-of-day lighting/rendering engineering (dynamic GI under a moving sun, sky and cloud rendering, precipitation and rain occlusion, the global param bus, performance), and persistence. References: BotW/TotK and Genshin (datamined), with RDR2/Horizon/Sea of Thieves for weather rendering, Stardew/RimWorld/Don't Starve for seasons, and Unity 6 / UE5 mappings. Use
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.