Public Agent Skills (agentskills.io spec) — portable SKILL.md capability packages

Craft reliable prompts for everyday tasks, tool agents, and software work.

A standard agent skill to adopt & sustain the harness-engineering paradigm on any project — scaffold the .harness/ loop, keep it alive, make it self-improving. 在任意项目落地并持续运转 harness engineering 范式的标准 skill。opencode / Claude Code / Codex / Trae.

Create software that is truly your own.

Skills modulares para asistentes de codificación por IA. Define reglas, flujos de trabajo y estándares técnicos mediante archivos Markdowm.

LLM prompt engineering for agent pipelines · Write, refine, and evaluate · 3 output styles: Anthropic · Google · OpenAI · Works with Claude Code, Cursor, Codex CLI, Gemini CLI & more · MIT

Agent Skills for professional GitHub repos - brief, structured, SEO-friendly READMEs that convert visitors into users, plus complete OSS scaffolding: community health files, templates, and discoverability best practices.

Governed, replay-tested agent skills for Codex-style maintainer workflows.

Loop-engineered goal mode for AI coding agents: visible previews, verification levels, checkpoints, delegation bounds, and terminal states.

One skill to bring Spec-Driven + Test-Driven Development (SDD+TDD) to any AI coding agent - OpenSpec x Superpowers bridge for Cursor, Claude Code, Codex, OpenCode, Gemini CLI and more

Trustworthy developer communication for AI-assisted open source.

Production-grade engineering skills for Cursor and Claude Code, full-stack architecture, TDD, security, Docker, queues, caching, and consistent UI.

Bilingual Claude Agent Skill + examples that route every LLM call (OpenAI, Claude, Gemini, DeepSeek, Grok, Perplexity) through one OpenAI-compatible Bifrost gateway for central logging. Python/PHP/Laravel/JS/curl.

Loopkit for Claude Code - guardrails, hooks, and two loop modes that won't let a task "finish" until it's actually done.

AI agent skills and evidence-backed workflows for understanding unfamiliar codebases, assessing change impact, and growing developers into confident maintainers.

Agent Skill for the Octopus Deploy CLI — works in Claude Code, Codex, Copilot, Gemini CLI and Cursor: releases, deployments, runbooks, task logs, plus the REST escape hatch for what the CLI has no command for

Make Chinese and mixed Chinese-English text sound more natural by removing formulaic GPT-style phrasing.

Read-only Interactive Brokers (IBKR) diagnostics as portable AI-agent skills: account health, positions, daily P&L, trade history, dividends, options Greeks, and a P0–P3 graded portfolio report. Never places orders.

AMOS 结构方程一条龙工作流 Agent Skill：数据清洗 → 8道达标预检验 → 自动绘图插件 → 脚本化分析 → 中介/调节 → APA撰写。含可运行的预检验脚本与 AMOS 自动绘图插件。

Human-led, evidence-gated full-cycle research execution for Codex and Claude Code · 人类主导的科研全流程辅助工具

Scan an AI agent skill for prompt injection, hidden Unicode, dangerous commands, and leaked secrets before you install it.

Evidence-first Agent Skill for reliable software engineering: reproduce, verify, root-cause, and document every change.

Agent skills for Clics: privacy-friendly, cookieless web analytics. MCP server and CLI for AI assistants.

Architect high-reliability, logic-driven AI agents for production.

A systems-engineering Agent Skill for evolving messy AI prototypes and agent workflows into structured, validated AI-native systems.

Agent-agnostic skill for writing lean, outcome-led GPT-5.6 prompts across coding agents.

Context-aware prompt refinement for Pi using GPT-5.6 Luna through a Codex subscription.

Security scanner for third-party AI agent-skill files (SKILL.md manifests, hooks, bundled scripts). CLI, library, MCP server, and GitHub Action, all reading the same 10 rule packs, including cross-skill privilege chaining and prompt-injection detection. Zero-auth, zero-config npx or pip scan.

面向 Agent 的研究生/本科生级源码学习 Skill：结合论文、真实代码证据与张量形状，按 Step 深入解析深度学习项目，并通过学习记录、反馈闭环和盲点审计持续复盘

Global persistent memory for OPENCODE sessions. Inspired by Claude Code's auto-memory — your agent remembers what it learns, across every session, globally.

Make coding agents prove their work with tests, builds, runtime checks, and honest evidence before they say “done.”

Portable, evaluated Agent Skills for product, business, and engineering workflows.

Cross-host skill for Codex, Claude Code, and Cursor that orchestrates parallel OpenCode agents with optional iTerm2 split panes.

Pre-mortems and postmortems for AI coding agents: 261 tagged real-world incidents + a 12-class agentic-AI failure model, shipped as Claude Code skills.

Turn a use case into a grounded agentic system design. An opinionated AI solutions architect as an agent skill.

Privacy-safe forward-deployed engineering skills for Claude Code, Codex, and AGENTS.md-compatible agents

Agent-neutral project orchestration skill from idea to delivery

Local-first capability router and safety gate for AI agent tools

Agent-agnostic software engineering skill for implementation, debugging, testing, refactoring, security boundaries, verification, and code review.

Skills that let a coding agent delegate work to another coding CLI, and keep itself in sync from git.

Agent Skills that design minimalist app and OSS package icons from a text brief or project context - symmetric logos on dark grey, Apple-style rounded-rectangle backgrounds.

Agentic variant of the idea-spark skill (microsoft/ResearchStudio): Claude Code sub-agents with WebSearch/WebFetch replace API connectors - no API keys required

Portable Agent Skill for delegating bounded AI-agent tasks to local CodeBuddy through ACP.

The agent context layer — converge messy multi-session AI-agent work into one lean, LIVING, spec-centered source of truth that doesn't rot. Organize docs by lifetime, keep the spec generative, and let an agent (or a human three months later) build from it.

Bilingual prompt compiler: one rough idea to Concise Universal + Advanced Max, without invented AI capabilities.

Track Agent Skill usage, estimated tokens, Rules/AGENTS.md bloat, and Task/sub-agent startup cost. Install: npx skills add wei63w/skill-monitor

Skills for early-stage startup founders.

Agent skill for Codex on Windows (installable for Claude Code too): triage sandbox startup failures by layer - CreateProcessAsUserW error 5 on the agent execution path, Git Bash/MSYS2 Win32 error 5 incompatibility, elevated setup-helper ACL failures, and config.toml permission tokens that can brick every session.

Agent plugin providing skills and MCP server configurations for Kuberoutectl scenarios.

One source. Every agent. Zero drift. | 跨 Agent 共享 Skill 管理系统 — 单一数据源，同步到 Claude Code / Codex / Hermes / OpenClaw

A coding-agent skill for safely planning, deploying, inspecting, and troubleshooting workloads on BytePlus Cloud.

Dynamically assemble dependency-aware AI teams for complex Codex tasks and decisions.

Extract Claude Fable 5's reasoning workflow into reusable Agent Skills so cheaper models like Opus 4.8 can inherit its behavior - no fine-tuning required.

AI-agent skill for building applications with WolverineFX framework

Safe repo-local starter template for Codex guidance, skills, read-only agents, conservative config, and no-key validation.

Lock down your code, refactor without fear: behavior-locking unit tests, mutation checks, and frozen-suite refactoring as portable Agent Skills for any coding agent

Official IntelliHelper UI agent plugin — Liquid Glass React components via MCP, skills, and slash commands

One automatic Codex coding entrance with right-sized rigor: lightweight everyday work, bounded multi-agent collaboration, and full delivery orchestration.

Give your AI agent a talking, lip-syncing face — 12-emotion particle face, in-browser Whisper (WebGPU), bring-your-own-agent bridge. MIT.

Evidence-driven engineering harness for autonomous coding agents