Agent Skills for prompt optimization - tune prompts for the latest frontier models like Fable 5 and GPT 5.6, write effective goals and success criteria, and engineer agentic loops that converge.

Community agent skills for safe, behavior-preserving refactoring (OpenCode, Claude Code, and any SKILL.md-compatible agent)

Agent Skill for GitHub Actions root-cause diagnosis, safe retries, bounty checks, AGENTS.md audits, and MCP schema drift via x402

Personal collection of reusable AI-agent skills for Codex and compatible runtimes

JEBI AGENT: clean handoff Skill for successor Codex chats

Generate static onboarding and architecture documentation for codebases

Skills for refactoring and optimizing React and Vite applications.

Markdown task-tracker system for Claude Code on long, multi-session academic projects

👁️ Give your coding agent eyes for 3D — MCP server that renders, inspects, and diagnoses meshes & Gaussian splats. Zero GPU/OpenGL dependency.

Official Agent Skills for SuperToken.

Measured prompt iteration with train/test dual-gate scoring and anti-overfit rails. Agent Skill (SKILL.md).

Drop-in OpenAI Codex skill that guides complete beginners from idea to working software through a perpetual loop workflow. 小白循环工作法向导（Codex 版）。Sister repo: loopwork-skill-claude-code

Claude Code skill that scaffolds the Ralph autonomous build loop: spec-interview to prd.json ledger to overnight self-driving builds

Agent Skill: weekly self-improvement loop for AI coding agents, analyzing their own local session logs (Claude Code, Codex, and more)

Claude Code skill + subagent for staged, evidence-gated rolling Splunk Enterprise upgrades across dev, test, pre-prod, and production.

A cross-platform Agent Skill for keeping observations, decisions, actions, and outcomes aligned with reality.

A factory for Claude Code skills. Nothing ships untested.

Claude Opus and Fable prompt engineering skill for focused frontend design, UI/UX, visual implementation and screenshot iteration

Portable evidence-based handoffs for AI agents

Local-first, chat-operated developer experiments, learning history, and approval-gated publishing.

Curated public Agent Skills with promotion evidence, portability checks, and repository-scoped integration guidance

Hardened Claude/Codex multi-agent build board with sealed contracts and deterministic gates

Moved to IchenDEV/prompt-optimizer-plugins. Historical Fable 5 v1.0.0 repository.

Agent skills for developing Spicetify extensions with AI coding agents: drive the live Spotify desktop client over CDP, and capture README/Marketplace screenshots.

A first-principles Vibe Coding prompt workflow for AI agents and chat models.

Private, local-first second brain for AI agents, with shared memory and governed skills across Claude Code, Codex, and your Tailnet.

Plugin toolkit for Claude Code and Codex.

Build a minimal, audited Agent Skill Stack for end-to-end goals, with safe discovery, conflict checks, and runtime routing.

用 4 行启发式编排 6 个 ECC Skills，通过三道确认门生成产品级 README 与 GitHub Landing Page。

Break past prompts. Engineer what matters. A vendor-neutral context engineering methodology and conformance kit.

Production-ready frontend Agent Skills for Codex, Claude Code & Cursor—covering React development, debugging, refactoring, code review, bilingual docs, and npx skills installation. 面向 Codex、Claude Code 与 Cursor 的生产级前端 Agent Skills，覆盖 React 开发、Bug 修复、重构、Code Review 与双语工程规范，支持 npx skills 安装。

Zero coding experience? Give Codex an AI delivery crew that turns ideas into working, tested, releasable software with guardrails, browser QA, rollback, and production closure.

Cat-grade engineering skills for coding agents. Spec → tickets → TDD in worktrees, plus repo hygiene. No slop.

A Codex skill for clear, evidence-based agent communication

A Codex skill for finding and reviewing GitHub projects by fit, maintenance, license, security, and integration cost.

Evidence-backed SharpEmu NID swarm, KawaiiDRA, Context Mode, and timestamped visual regression tooling

Nine production Claude Code skills and one command pack: repo security scanning, LLM-output validators, agent-fleet guardrails, credential hygiene. Deterministic cores, agentic edges.

Custom agent skills for GitHub Copilot & Claude Code, from Microsoft-ecosystem dev experience (Azure, .NET, M365, Graph).

Human-led judgment and agent-executed harness engineering skills

余生姜在真实业务中持续使用、测试和迭代的 AI Skills

Flame-graph-style web performance auditing agent skill

Author portable plugins for coding-agent hosts.

A cross-compatible Agent Skill for delegating bounded coding tasks from Codex or Claude Code to a supervised Claude Code Sonnet worker.

Agent skill: SwiftData test patterns (ModelContainer fixtures, mock repos, @ModelActor tests, Decimal money assertions, import-hash dedup)

Production-minded Agent Skills for responsive web design, root-cause refactoring, and maintainable app architecture.

A read-only Codex skill for finding duplicate, misplaced, risky, and structurally inconsistent files across Git repositories.

Turn ASCII architecture diagrams into native, editable Monodraw documents with Codex or Claude Code.

Claude to Codex Migrator: convert Claude skills, plugins, and assistant packages into validated Codex packages.

Agent Skill: reversible PII tokenization — masks personal data before the LLM sees it, restores it in the final answer (SKILL.md, works in Claude Code and any agent)

Automate your social media with Handled Agent Skills: guided Facebook onboarding, posting rhythm, approvals, and Social Connect workflows. More platforms coming soon.

A portable, evidence-led forensic audit skill for Codex, Claude Code, OpenClaw, and Gemini.

Local YouTube transcription skill for AI agents — Whisper-always, self-improving vocabulary memory. No captions, no cloud.

Cohort Method — portable Agent Skill for multi-agent creative panels, UX swarms, build cohorts, and cross-QC (10 agents / 5 cohorts).

Reusable coding-agent skills for feature delivery, review, and software engineering workflows

Reusable agent skills from Dead Pixel Design, starting with an agent-agnostic Codex prompt compiler.

Agent skill for Claude Code and Codex that prevents silent no-op failures when delegating to subagents — mandatory prompt clauses, artifact verification, resume-based recovery, and ledger-as-spec parallel delegation.

Agent evals on autopilot: find quality bugs in your AI agent, ship a targeted fix, and prove it on a held-out set. Zero-dependency Agent Skill + CLI.