Browse all resources

A Skills-driven software development methodology covering discovery, planning, implementation, and verification.

A collection of 25 installable Agent Skills maintained by Matt Pocock for real software engineering, including discovery, specification, TDD, debugging, code review, domain modeling, tickets, and handoffs.

Anthropic's public Agent Skills catalog covering document, design, development, and content workflows.

A community-built prompt discovery and collection project with a website, CLI, plugins, and MCP support.

An archive of system prompts, tool definitions, and model information organized by AI product.

The official collection of reference servers and example implementations maintained by the Model Context Protocol project.

An open learning resource covering prompting, context engineering, RAG, and AI agents.

A complete meta-framework for multi-agent orchestration, memory, MCP, and adaptive workflows.

Provides coding agents with current library documentation and version-aware code examples.

A multi-harness plugin marketplace combining specialized agents, Skills, and workflows.

Anthropic's official interactive course covering prompt structure, few-shot examples, tool use, and retrieval.

Lets agents navigate websites, read pages, and perform actions through structured browser interaction.

Lets AI agents access GitHub repositories, issues, pull requests, and Actions through controlled tools.

Frontend, React, deployment, and web design Skills published by Vercel Labs.

Serena is an open-source MCP toolkit for coding agents. It uses language servers or an optional JetBrains backend to provide symbol search, reference lookup, rename, and structured code editing, with official setup for Codex CLI and App.

OpenAI's curated Skills catalog for Codex, including research, deployment, document, and development tools.

Convert technical books, PDFs, document folders, and source collections into structured Agent Skills that Codex, Claude Code, GitHub Copilot CLI, and Amp can load on demand.

Google's official Agent Skills catalog provides selectively installable guidance, references, scripts, and plugin entry points for Google Cloud, Ads, Analytics, and developer platforms.

A professional collection of Skills and plugins for security research, code auditing, and vulnerability analysis.

A Tencent Zhuque Lab Skill that statically reads Skill definitions, scripts, manifests, and configuration to identify permission abuse, privacy access, dangerous operations, hardcoded credentials, and prompt override behavior without executing the scanned code.

A Tencent Zhuque Lab Skill for authorized assessment of AI agents, MCP servers, Skills, code, and AI infrastructure, with threat modeling, harmless proof, and evidence-backed reporting.

A unified virtual filesystem that connects DSH and other agents to multiple data sources.

An Agent Skill for Chinese writing and revision that checks whether the source material is sufficient, separates factual and fictional workflows, and reviews drafts for repetition, canned reversals, report-like phrasing, and flat sentence rhythm.

OpenAI's self-hosted client for connecting private or localhost MCP servers to ChatGPT, Codex, the Responses API, and AgentKit through Secure MCP Tunnel without exposing the MCP listener to the public internet.

A two-turn prompt that delays the final answer, asks the model to identify assumptions, missing information, and a common mistake, and then asks one key question before responding.

An engineering workflow pack combining agents, skills, hooks, commands, and rules for planning, implementation, testing, security, and continuous learning.

MarkItDown-MCP is the MCP server in Microsoft's MarkItDown repository. Its `convert_to_markdown(uri)` tool converts local files, web pages, or data URIs into Markdown for LLM workflows.

A library of specialist agent personas organized by role and task, with conversion and installation scripts for Codex and other coding agents.

GitHub's official open-source spec-driven development toolkit, connecting requirements, clarification, planning, tasks, implementation, and convergence checks through reviewable prompt commands.

A cross-platform desktop configuration manager for switching providers, MCP servers, prompts, and skills across Claude Code, Codex, Gemini CLI, and other tools.

Ponytail gives Codex, Claude Code, OpenCode, and other coding agents a minimal-implementation ruleset, six companion Skills, and lifecycle hooks. It prioritizes existing code, standard libraries, and native platform features without removing security, validation, error handling, or accessibility safeguards.

UI/UX Pro Max is an installable design-intelligence Skill for Codex, Claude Code, Cursor, and other AI coding assistants. It uses locally searchable data to generate product design systems and provide style, color, typography, interaction, accessibility, and implementation guidance across 22 stacks.

Microsoft's open generative AI course with 21 lessons on prompt engineering, text apps, chat, search, image generation, and agent foundations.

An Agent Skill that turns code, documentation, SQL, configuration, PDFs, and media into a queryable knowledge graph, using deterministic local AST extraction for code and provenance labels for relationships.

A hook-based memory plugin that records coding sessions and recalls relevant context through a local worker, database, and search interface.

A suite of 24 engineering skills and eight lifecycle commands for AI coding agents, covering specification, planning, implementation, testing, review, and shipping.

Agent Reach gives command-capable AI agents workflows for reading and searching multiple platforms, selecting upstream tools, checking dependencies, and diagnosing channels. Some social sources require authenticated sessions or additional setup.

The example collection from the official OpenAI Cookbook, covering prompt design, structured outputs, tool use, evaluation, and production patterns.

Microsoft's open 18-lesson AI agent course covers design patterns, tool use, agentic RAG, multi-agent systems, context, memory, deployment, and security with Python and .NET examples.

A frontend design Skill for AI coding agents with 23 interface workflows, live browser iteration, and 61 deterministic detector rules, plus local hooks that surface findings during edits.

last30days is an Agent Skill that researches recent discussion across Reddit, Hacker News, GitHub, Polymarket, X, YouTube, and other sources, then organizes the last 30 days by engagement signals.

Anthropic's official collection of Claude examples and guides for tool use, the Agent SDK, extended thinking, multimodal work, evaluation, and prompting patterns.

An MCP server that connects agents to Chrome DevTools for page debugging, performance analysis, network inspection, and browser automation.

Obsidian Skills teaches AI assistants the file formats used for notes, wikilinks and visual canvases. Existing Obsidian users can start with file-based skills through manual copying, without programming. Command-line control and web extraction are separate optional capabilities.

A large community Skill catalog and local discovery toolkit with a CLI, MCP server, plugins, and Workbench.

A collection of reusable Markdown Prompt Patterns, strategies, and a Go CLI for building inspectable, customizable AI workflows.

HeyGen's open-source video framework and Agent Skills suite for authoring, previewing, and rendering deterministic MP4 videos from HTML, CSS, media, and seekable animation.

Diagram Design is a technical-diagram skill for Codex, Claude Code, Pi, and other agent hosts. It creates self-contained HTML with inline SVG and redraws draw.io, Mermaid, and Excalidraw sources.

Humanizer gives an AI assistant editing rules for reducing vague claims, repetition and chatbot filler. Upload the repository ZIP in Claude Desktop or use a documented agent installation route, then provide a draft and an optional sample of your own writing.

A GitHub-hosted community collection of installable Copilot plugins plus agents, skills, instructions, hooks, workflows, and copy-ready guidance.

Claude Code Plugins Directory is Anthropic's official marketplace repository for discovering and installing internal and partner plugins through Claude Code.

More than 160 agent skills for scientific work, spanning research databases, analysis tools, and scientific writing workflows.

A web, desktop, and browser-extension tool for optimizing, testing, evaluating, and saving prompts, with self-hosting options.

OpenAI's official Codex plugin for Claude Code provides eight slash commands for read-only code review, adversarial review, task delegation, rescue work, background job management, and session transfer while reusing the local Codex CLI's authentication and configuration.

A collection of Agent Skills bundled with Google Workspace CLI that helps agents operate Gmail, Drive, Calendar, Sheets, Docs, and other Workspace services through gws, with cross-service workflows and safety rules.

A discovery catalog of more than a thousand agent skills organized by source and use case, covering official teams and community projects.

A CLI, template, and component collection for configuring, extending, and monitoring Claude Code.

A DeepSeek Harness plugin that connects OpenViking's context database and retrieves relevant long-term memories before task steps.

A Python MCP application framework for building servers, clients, and interactive apps through one API, with stdio and Streamable HTTP transports.

A skill that keeps plans, progress, and findings in persistent files so agents lose less context during long tasks and can recover after interruptions.

Claude Skills is a cross-platform collection of hundreds of skills, agents, commands, and scripts spanning engineering, product, marketing, research, compliance, finance, and daily productivity.

Every's open-source agent workflow plugin connects discovery, planning, implementation, debugging, review, delivery, and captured learnings through 33 skills, with native installation paths for Claude Code, Codex, Cursor, and other hosts.

A CLI and library for versioned prompt and model-output evaluations, plus red-team testing for LLM applications, agents, and RAG workflows.

The open Agent Skills format and documentation, defining how SKILL.md, scripts, references, and assets form a portable capability package.

A configuration framework that enhances Claude Code with specialized commands, personas, and development methods.

Knowledge Work Plugins is Anthropic's public marketplace of job-oriented Claude extensions. Choose an office or sales workflow in Cowork, start with information you supply, and decide separately whether to connect company systems.

Browser Use's conversational video editing Skill for cutting pauses, filler, and retakes, adding subtitles and color treatment, and reviewing rendered cuts from transcripts and on-demand timeline views.

Hindsight memory integration for DeepSeek Harness, Codex, and Claude Code, using hooks and MCP to retain and retrieve project context.

Google's open-source database MCP server and custom-tool framework connects many databases to MCP clients while letting operators constrain agent actions through prebuilt or parameterized tools.

AI Berkshire packages value-investing research into 20 callable Skills for company research, earnings review, industry screening, portfolio review, and data validation, with separate install paths for Claude Code and Codex.

A DeepSeek Harness integration that runs Archify against a codebase and produces structured JSON plus an interactive HTML architecture map.

img2threejs lets Claude Code, Codex, or OpenCode rebuild an object or character reference as animation-ready procedural Three.js code through staged specs, renders, and review gates.

LangGPT is a method for writing reusable prompts with separate role, goal, rules and workflow sections. Copy its Markdown template into a chat interface and adapt it for writing, learning or information tasks without installing software.

Loop Engineering provides reusable prompt patterns, scaffolds, and CLI tools for running budgeted, human-gated repository loops such as daily triage, PR babysitting, and CI cleanup with Codex, Claude Code, Grok, and OpenCode.

Hugging Face Skills is the official Agent Skills repository for using hf-cli and task-specific skills to search models, manage datasets and Spaces, evaluate models, and plan training workflows.

MCP Inspector is the official Model Context Protocol debugging tool. Its Web, CLI, and TUI clients connect to MCP servers and inspect tools, resources, prompts, transports, authentication, and MCP Apps.

An AWS Labs MCP server that lets agents search, read, and retrieve recommendations from official AWS documentation.

China Patent Skill organizes invention mining, disclosures, application drafts, patent reading, and search into separate workflows, with Word documents, diagrams, and optional Obsidian notes. Formal use requires professional review.

A compact prompt engineering guide covering model limitations, prompt risks, and practical strategies.

A local tool server for desktop MCP clients, exposing file read/write and search, surgical edits, terminal commands, process management, and structured-data analysis over STDIO. Its broad host permissions require narrow directory scope and container or VM isolation for sensitive work.

Cangjie Skill applies the RIA-TV++ pipeline to turn books, transcripts, podcasts, courses, and interviews into verified, composable, pressure-tested Agent Skills.

Provides web scraping, search, content extraction, and batch processing tools to MCP clients.

video-shotcraft is a product-video production Skill for Claude Code and Codex, with 152 shot recipe cards, 209 motion previews, Remotion implementations, the Ink Press template, and sound-design assets.

An MCP server and standalone CLI that give coding agents access to Xcode builds, tests, simulators, devices, debugging, and UI automation.

A hand-curated index of prompt and context engineering resources, covering papers, tools, models, benchmarks, courses, and security material.

Tencent’s DeepSeek Harness plugin for BrowserSkill exposes page inspection, interaction, screenshots and isolated browser sessions through the local bsk CLI and browser extension.

Peter Yang's writing Skill identifies and edits more than 20 named AI-slop patterns while requiring minimal changes and preservation of the writer's vocabulary, cadence, humor, and imperfections.

A DeepSeek Harness integration for the Ouroboros self-improving agent runtime, with DSH or alternative model backends for iterative optimization loops.

Add a task board, Git graph, right-side workbench, mobile remote access, SSH, image understanding, and a skin center to DeepSeek Harness Web.

Exa MCP Server adds web search and public-page fetching to AI assistants. Its hosted endpoint can be connected through compatible remote-MCP clients without running a server. Anonymous use is rate-limited, while higher limits and research tools have separate authentication and cost boundaries.

IP as Logo is an open-format Agent Skill that proposes three character directions, waits for approval, and then generates six independent square candidates. It uses bold silhouettes, three semantic colors, lower-corner composition, and a strict detail budget to keep characters simple and recognizable.

Notion's official MCP server, using an internal integration token to let agents read and work with authorized pages and databases.

Callstack's cross-platform app automation toolkit drives mobile, TV, web, and desktop apps through a stdio MCP server, CLI, and typed TypeScript API while collecting screenshots, logs, network, and performance evidence.

Agent Skills maintained by the Remotion team for creating React video projects, authoring compositions, previewing, rendering, captions, map animation, Studio interactivity, and upgrades.

A suite of image, video, and audio generation Skills for coding agents, combining the MuAPI CLI, production knowledge, workflow recipes, and an MCP server. Generation requires a MuAPI key.

A local plugin for Claude Code or Codex that assigns optional model providers to research, design, implementation, and review roles, then surfaces disagreement and consensus without dispatching external providers by default.

Start the first request with Minimal-aligned tools, then restore Standard tools and context to study first-request anchoring and experimental workflows.

Bytebase’s open-source database MCP server connects six database engines and exposes object discovery and SQL execution over stdio or HTTP.

Convert pasted images into structured OCR, layout, entity, and relationship evidence so text-only DeepSeek or GLM models can work with screenshots and document images.

A bilingual prompt-writing Skill for Jimeng Seedance 2.0 covering multimodal reference syntax, camera language, timed segments, sound direction, and templates for ads, dramas, music videos, and education.

An official Grafana MCP server that lets compatible AI clients discover dashboards and data sources, query Prometheus metrics and Loki logs, and work with alerts, incidents, and OnCall flows within existing Grafana permissions, with read-only, query-free, and network-hardening controls.

Add file editing and previews, real terminals, Git changes and side conversations to DeepSeek Harness Web, with right and bottom panels, floating windows and third-party extensions.

A Codex Marketplace video editing plugin maintained by Chengfeng. Six task-facing Skills orchestrate talking-head cuts, subtitles, visuals, export, safe bug reports, and trusted update checks through a managed runtime.

Claude Plugins Community is Anthropic's read-only community marketplace mirror for Claude Code and Claude Cowork. It indexes third-party plugins that passed automated security scanning and distribution review, with most external sources pinned to exact commits.

Supabase's official MCP server, connecting agents to project configuration, databases, and developer tools through hosted OAuth or a local endpoint.

A pre-execution security layer that checks suspicious URLs, pipe-to-shell commands, terminal injection, data exfiltration, and malicious agent configuration through shell hooks, host hooks, or an MCP gateway.

sepia handles structural and stylistic AI-writing patterns by text type: fiction gets narrative repair before polishing, professional documents receive venue rules, and users choose drafting, diagnosis, minimal edits, or full rewriting.

MCPHub is a self-hosted gateway between AI clients and multiple MCP servers, combining endpoint aggregation, group routing, access control, credential isolation, runtime logs, and optional vector-based smart routing.

SRT Whiteboard Animation parses subtitles, plans scenes, coordinates line art and semantic regions, then renders continuous ink-and-color whiteboard animation to MP4.

Add a full-screen terminal interface to DeepSeek Harness with streaming thoughts, tool cards, context progress, TPS, session actions, and rollback.

A native Go MCP server that exposes Kubernetes and OpenShift resources, pod diagnostics, events, Helm operations, and optional ecosystem toolsets, with controls for read-only access, denied resources, and RBAC-scoped identities.

Azure DevOps MCP Server lets MCP-capable agents read and operate on Azure DevOps projects, work items, repositories, pipelines, test plans, wikis, and Advanced Security data.

Writing DNA Distiller analyzes at least 20 complete articles across language, structure, topics, sources, cognitive frames, and visuals, producing four layered documents and Writing-DNA.md for an agent to reread before drafting.

A Timescale-maintained PostgreSQL agent extension combining local best-practice skills with hosted MCP documentation search for PostgreSQL, TimescaleDB, and PostGIS.

Stripe MCP Server lets agents search Stripe documentation and manage customers, products, prices, payment links, invoices, refunds, subscriptions, and disputes within Restricted API Key permissions.

Browse, search, install, update, uninstall, and switch community plugins and themes inside DSH settings, with backup and restore support.

Awesome Seedance collects source-traced Seedance 2.5/2.0 video prompts, 25 copy-ready templates, cross-model retest records, and 59 installable Skills to help creators turn a shot intent into a structured video prompt.

Provide maid-atelier and orca-link Whale Girl themes for DeepSeek Harness Web, with light and dark previews and local-path installation.

A pre-command hook that checks shell operations and blocks or warns about commands that could cause data loss.

A Microsoft HVE-maintained GitHub Copilot workflow pack that combines agents, prompts, instructions, and skills for research, planning, implementation, review, and security checks.

Review code, architecture, technical debt, and tests through checks grounded in twelve classic software-engineering books, with cited findings and severity labels.

Inspect context occupancy and composition, request trends, full model-visible content, file activity, and agent relationships inside the DeepSeek Harness Web UI.

Connect Feishu, WeChat, DingTalk, WeCom, QQ, Slack, Telegram, Discord and WhatsApp to DeepSeek Harness for continuing sessions, approving tool calls and exchanging files.

MongoDB's official MCP server for inspecting schemas, running queries, and assisting with database operations, including a read-only mode.

Apply baselines, evidence verification, and drift checks to debugging, architecture changes, and long tasks, with native DeepSeek Harness profile installation.

A Microsoft 365 and Power Platform community collection of copyable prompts, Copilot Studio agent instructions, and GitHub Copilot skills.

A HashiCorp-maintained collection of 20 Agent Skills covering Terraform configuration, provider development and testing, policy, plus Packer workflows for AWS, Azure, Windows images, and HCP Registry.

Power Platform Skills is Microsoft's official agent plugin marketplace for Power Pages, Power Apps, Power Automate, and related development workflows in Claude Code and GitHub Copilot CLI.

Provide agent-callable tools for image Q&A, long-screenshot OCR, crop, pixel and color checks, UI reconstruction, and visual diffs.

Turn the current session into a captain that creates durable subagents, dependency-aware tasks, and messages, with a live team activity panel.

A local workflow plugin that connects durable specs, isolated implementation, cross-model review, evidence receipts, pull requests, and landing across Claude Code, Codex, Cursor, Droid, Grok Build, and OpenCode.

A MicrosoftDocs-maintained collection of Azure Agent Skills that turns decisions, configuration, security, troubleshooting, and deployment knowledge from Microsoft Learn into standard SKILL.md resources for Codex, Claude Code, GitHub Copilot, and other coding assistants.

Turns WeChat and Xiaohongshu articles into 3:4 vertical cover prompts through three rounds of questions about style, source material, and visual details.

An open-source product management skill library from Product on Purpose, using 68 composable skills, 12 workflows, and six sub-agent definitions to structure outputs from research and problem framing through PRDs, measurement, and retrospectives, with installation paths for Claude Code and other agents.

Add parody 2005-style portal ads, fake games, popups, and plugin promotion slots to DeepSeek Harness Web for entertainment and demos.

Search workspace files and directories in the DeepSeek Harness Web composer and attach structured workspace-relative path references to prompts.

Yingzao is an Agent Skill for Claude Code, Codex, and other hosts that support Skills and image-generation tools. It inspects real photos and factual boundaries, then combines a dominant reference, Chinese display-type design, and a sparse layout guide in one integrated image edit, with an optional 3×3 video storyboard after user approval.

MapStage is an open-source Skill for coding agents with a browser-based tuner for map and globe projections, satellite imagery, hypsometric relief, hillshade, 3D terrain, region isolation, and JSON preset export.

Display an iOS Simulator or USB-connected iPhone beside a DeepSeek Harness conversation, with device control, Xcode builds, UI inspection, logs and SwiftUI previews.

LiveCanvas is an open-source Skill for Codex, Claude Code, and other AI coding assistants. It researches and checks source material, builds a Remotion animation with a designed cover, scenes, images, emoji, timelines, or charts, and can package Apple Live Photo resources on a supported macOS environment.

Tencent's official WeChat Reading Agent Skills use a unified API Gateway for book search, bookshelf access, reading progress, notes, reading statistics, public reviews, and personalized recommendations.

Connect Android emulators or USB phones to DeepSeek Harness through adb, view a live stream, interact with the UI, build Gradle projects and inspect logcat, processes and memory.

OJO Design Skills gives coding agents two UI/UX routes: proven systems for utility products and distinct directions for brand-led experiences, followed by design tokens, component states, and motion specifications.

A visible browser shared by a person and a DeepSeek Harness agent, with 34 browser_* tools, task-isolated windows, persistent sign-in state, forms, screenshots, downloads, and human takeover.

Runpod MCP server is the official control-plane MCP server for managing Pods, Serverless endpoints, templates, network volumes, and related cloud resources over hosted HTTP or local stdio.

Let the agent search GitHub's dsh-plugin topic from the session, rerank results by stars, and return descriptions with install commands.

An Agent Skill and standalone Python CLI that uses ExifTool to batch-remove supported C2PA, Content Credentials, OpenAI/gpt-image provenance, and common EXIF, XMP, IPTC, comment, and software metadata while preserving pixel data, ICC color profiles, and modification times by default.

Open the WeShop infinite canvas inside DeepSeek Harness so agents can generate and organize product visuals while synchronizing work between the session and canvas.

A community plugin that brings project and user Claude Code skills, commands, rules, hooks, agents, MCP servers, plugins, and sessions into DeepSeek Harness.

DA Motion Sticker Skill turns one character reference and nine reactions into nine separate transparent animated GIFs, with 36 styles, Codex key-pose and external AI-video routes, local keying, validation, and ZIP packaging.

A local MCP server that connects AI clients to Z-Library book search, download links, and account quotas. It uses your own account over stdio and can save selected books locally when a download directory is configured.

Inspect dangerous shell, SQL, and file-write actions before execution, redact common credentials from results, and provide a read-only security review workflow.

Uses six to nine main questions to map health, work, play, love, values, and energy, separates gravity problems, and creates three five-year paths with prototypes.

Uses up to ten main questions about early interests, unconscious competence, energy patterns, and envy, then builds a personal talent manual from the user's evidence.

Finds three assumptions behind a decision, selects the one most likely to change the conclusion, and designs a small experiment with metrics and stop or continue criteria.

Builds the strongest case for both directions, compares conditions, benefits, risks, and hard objections, and asks one question most likely to change the conclusion.

Removes domain jargon to identify the problem structure, searches history and at least three distant fields for analogous mechanisms, and ends with a reversible experiment.

Reduces a problem to unavoidable facts, unverified assumptions, the real goal, and practical constraints, then derives a feasible path without relying on convention.

Selects three genuinely complementary professional perspectives, compares their paths and risks, and uses cross-examination to identify consensus, disagreement, and assumptions.

Separates a claim into verifiable facts, inferred conclusions, and value judgments, checks sources and context, and then audits the reasoning chain.

Studies how a subject developed over time, compares it with relevant peers using consistent dimensions, and connects historical capabilities and constraints to future paths.

Reverse-engineers a strong product, page, process, or finished work into its audience, structure, key decisions, definition of done, reusable principles, and a practice exercise.

Explains one concept in a beginner-friendly form and a precise professional form, then maps the terms, flags misconceptions, and asks comprehension questions.

A multi-turn prompt that delays advice, asks one decision-changing question at a time, separates facts from interpretations and goals, and ends with a more precise question.