Writing & Editing

Research the person before you write to them. A Claude skill and agent contract that refuses to draft outreach until a verified research hook exists.

Task complexity routing, phase planning, and compact handoffs for Codex and AI agents.

Offline, local-only context migration skills across 9 coding-agent sources × 9 destination hosts (registry-derived 81 cells). Stdlib-only; inert handoff for fresh sessions, not live restore.

Agent skill for Claude Code and Codex: create PRs from a temporary git worktree off origin's default branch when the main checkout is dirty, behind, or shared with other agents - with merge-mode-aware cleanup guards.

A predictable CLI toolbox for coding agents — fully local ML: transcribe speech (Whisper, GigaAM, Vosk), read text aloud in a preset or cloned voice (Qwen3-TTS, ESpeech/F5-TTS), cut silence by voice activity, restore punctuation and casing, split transcripts into paragraphs; plus web feeds. JSON-first, stable flags, meaningful exit codes.

Evidence-gated Planner-Executor-Verifier loops for long-running Codex goals, with fresh role contexts and OS-temp-only run state.

Standalone Codex Skill for tracking job applications from NetEase 163 Mail

Portable Agent Skill for active-chat-model PDF translation with resumable batches and layout-preserving rendering

Codex skill for evidence-grounded Korean meeting minutes and verified HWPX generation

A local, faithful, searchable warehouse of your agentic-coding history.

A semantic repair loop that actually fixes code. State the success condition in prose, not just as an exit status, and the loop refuses the repairs that make a command pass without fixing the defect.

A game-agnostic AI agent skill for safely discovering, evaluating, subscribing to, and verifying Steam Workshop content.

Skill para iniciar projetos web, iOS e macOS com o setup e as preferências do Lucas

A game-agnostic AI agent skill for evidence-based Steam mod conflict auditing.

Agent Skill for public-trend-based book, movie and TV recommendations.

An MIT-licensed Codex skill for governing efficient multi-task waves with clear ownership, verification, integration, and cleanup.

Personal AI development toolkit — skills, hooks, and verification gates that work in both Claude Code and Cursor. Installed once globally; product repositories stay untouched.

Two-role workflow for AI coding agents (Claude Code, Codex) over GitHub Issues, Linear or Trello: an analyst files evidence-backed issues, a dev claims and ships them. Race-safe claiming, per-run attribution, pluggable domain rule books.

Automated Systematic Literature Review (SLR) pipeline that identifies relevant studies from academic databases, downloads accessible papers, conducts multi-stage screening, and generates a curated, reproducible set of papers ready for synthesis from a research question, with agent-driven execution and human-in-the-loop where needed.

アジャイル/スクラムの規律を、AIとの協働に適用する — Claude Code / Agent Skills を活用したAI駆動開発（AIDD）の実践ガイド

🧩 Curated, auditable agent skills for knowledge, mastery, career, and cross-system orchestration.

Autonomous multi-agent orchestrator skill: fans out sub-agents to explore, diagnose, fix, review and verify a codebase — cross-checked, ledger-backed, doesn't stop until it's production-ready

江南 × 烽火戏诸侯文风蒸馏 Agent Skills（Claude Code/Codex/Cursor/Trae/Antigravity 通用）｜1465万字语料实证 + 自动风格审计 + 独立盲测 95/100·A

Two agents spar over one codebase until every review finding is reconciled.