개인용 에이전트 스킬 모음집

Agent Skills for building and operating LLM councils - multi-model deliberation with anonymized peer review, robust aggregation, calibrated escalation, and defenses against correlated failure.

Two staff-level agent skills (Claude Code / Copilot CLI / Codex). interview-me-senior grills and scores your decisions; review-me-senior is a review copilot with auto-detected best-practice packs (C#/.NET, Blazor/Fluxor) that learns your conventions.

Skills, subagents, commands, and tech conventions for opencode — one harness, no fragmentation.

The project board that writes itself while you build.

An Agent Skill for writing polished, simple, evidence-driven GitHub project READMEs

Global dlt-first AI skill for Meta Ads and Google Ads PostgreSQL warehouses

Agent Skill nghiên cứu và luận giải Tử Vi Đẩu Số Việt Nam có đối chiếu, phản chứng và nghiệm lý

Learn by doing: the AI breaks work into small steps and reviews your code, but never writes it for you.

A skill that forces the clearest, shortest nonfiction prose that survives a tired reader.

Codex UI anti-slop skill for visual contracts, DESIGN.md tokens, and screenshot QA

Keel turns agent planning into an auditable contract. It stores state in an append-only .keel/ledger.jsonl, requires every phase to carry an executable acceptance check, and supports recovery, stop-gate verification, multi-agent coordination, and research experiment deduplication.

开源 AI 财报分析 Skill：审阅业绩预期差、盈利质量、现金流、管理层指引和研究论点变化。

Git-native persistent memory for AI coding agents via MCP. Store structured project knowledge as reviewable JSON files instead of a database.

Open-source Codex 总指挥 / Project Commander with task-window employees, completion watchdog, idle-capacity audits, automatic re-dispatch, and Token routing

Agent-oriented workflows and validators for developing Hotwire Native apps

Agent Skill: teach your AI agent to build and run nanoodle graphs — multi-model media pipelines as one tool

A workshop where skills for LLM agents are engineered, not just written: developed test-first against baseline agent behavior, hardened by adversarial review, and required to earn their keep with a logged field-win record.

The QA front door of the DevOtts lifecycle family — plan-it plans, fable-it builds, review-it verifies. Runs the plan-phase Test Contract against the build and enforces an 11-rule gate catalog that makes false-VERIFIED claims un-shippable.

Claude Code plugin: agent skills that turn a PlayStation play-history export into taste reads, backlog verdicts, and wishlist calls. Built around awesome-psnstats.

Redraw PowerPoint decks into visually coherent, editable PPTX files with Codex.

Agent skills for Laravel and PHP development. Installable via npx skills@latest add ceilidhboy/skills

Agent skills from the Ottto team. First up: the landing lander - an agent-run PR merge queue.

Five open agent skills for sharper literature reviews: question refinement, review modules, method comparison, contradiction mapping, and research gap discovery.

17 executable AI agent skills for Cocos Creator 3.x debugging

Free LinkedIn & content skills for AI agents — by Liftli (liftli.ai). Install: npx skills add liftli-ai/skills

An approval-first agent skill for exploring UI directions, validating interactive HTML prototypes, documenting approved designs in Figma, and handing work off to implementation.

Forge AI advisor councils from real experts' public thinking — methodology + templates for Claude Code, hermes-agent, and OpenClaw

Agent Skill for auditable experiment reports from CSV and JSONL training logs.

Codex-native frontend skills, custom agents, profiles, and an evidence-backed evaluation harness for React, Next.js, Vite, and Astro.

Agent Skill for PSNR/SSIM image benchmarks with per-image scores and worst-case reports.

Agent skills for Cursor: Confluence/Jira → domain briefs → story risk & INVEST splits and test case Design. Offline DEMO-1 — no Atlassian needed to try.

Source-aware Western astrology Agent Skill separating astronomy, calendars, and symbolic traditions

Source-aware Tarot Agent Skill with 78-card study data and reproducible reflective draws

A skill for Anthropic's ecosystem that refines underspecified prompts into structured, surface-correct, ready-to-use instructions for the Claude ecosystem.

Goal-first Codex production router for video, web, ads, images, audio, documents, and engineering workflows

A repo of skills I use in my current personal workflows. I keep updating as I create new skills

Roblox Studio skill and local MCP bridge for Claude Code, Codex, and other coding agents.

Agent skill: multi-agent adversarial code audit — run every installed AI coding agent CLI (Claude, Codex, Gemini, OpenCode, Kimi, Cursor…) in parallel on one brief, verify every claim against the code, score who was right/wrong/hallucinating

Agent skill: a task is done only when verified and actually works. Triages scale, recons your repo, ships verified code or code + an honest run plan. For Claude Code, Codex, Cursor & other agents.

Codex skills for writing idiomatic Rust with curated guidelines and strict Clippy checks.

Claude Code / Codex / Antigravity 用の Agent Skills 集

Agent Skills to delegate questions, reviews, and implementation to Google Antigravity CLI (agy) from Claude Code / Codex

pkgdown for AI agent skills: generate a Quarto documentation site from a skill collection (R package)

Scaffold, lint and audit Claude Code / Agent Skills. Zero runtime dependencies.

Codex and Claude skill for Stanford STORM deep research and cited report generation

Verified Markdown checkpoints for resumable long-running agent work.

Cross-model second opinions for coding agents — skills that let Claude Code consult rival-model CLIs (Codex, Grok, OpenCode), solo or as a blind three-model panel.

A Claude Code skill and hook that catches silent tool failures where exit code 0 masks a real error — it verifies a command's actual output (pytest under xdist, linters in report mode, crashed background jobs, placeholder writes) instead of trusting the exit code, and loops back to fix them.

Cross-platform tool-routing architecture for Codex, Claude Code and zcode. Agent 工具路由、生命周期管理与安全回滚。

Installable skills, gates, hooks and templates for deliberate human-agent work — drops into your existing Claude Code or Codex setup.

Journal Impact Factor lookup skill — bundled JCR CSV + OpenAlex API fallback. Agent-native CLI with stable JSON envelope, fuzzy matching, and batch mode.

AI hiring-signal tracker for investment research: turn official job-posting APIs into a weekly time series of AI-investment signals. Python stdlib only. Agent Skills compatible.

Compile distributed agent skills into one provenance-aware router skill

Public AI agent skills mirrored from jsolly/dotagents (allowlisted)

System design knowledge for AI coding agents — a Claude Code plugin. 1 skill, 10 commands, 14 self-contained reference files. State the constraint, pick the option, name the tradeoff.

Seed fake-but-production-like demo databases with media, demo accounts, and a manifest.

Codex skills for Effect V4 tooling

Open-source Agent Skill for standards-based professional document creation (ISO/IEEE/APA/PDF-A, i18n/RTL).

Claude skill: chuẩn bị hồ sơ đăng ký bản quyền phần mềm (Vietnam software copyright) — chụp UI + mã nguồn, che mờ dữ liệu, lắp ráp .md/.docx/PDF. Generic cho mọi dự án.