skills plugins

by river-review maintainers

Turn review into an organizational judgment asset. Review Judgment as Code for AI-assisted development, delivered as a capability pack for Codex: your team's review criteria as versioned, repo-owned skills (a Skill Registry), driven by a review agent that runs perspective-based reviewer roles (code, security, performance, architecture, testing, adversarial) and verifies findings against the diff so humans keep the high-risk judgment.

by Jesse Vincent, forked by brunob54

Structured workflow skills for Codex — systematic debugging, TDD, brainstorming, code review, and 30+ expert workflows. Skills work on all platforms. Lifecycle hooks (routing, safety, reminders) require separate setup on macOS/Linux.

by Allen Eubank

Reusable planning, handoff, verification, and workflow skills for agents.

by Daniel Luca

Orchestrator skills for PR review, design review, issue triage, simplification, batch merging, codebase mapping, and installing periodic maintenance agents.

by akbun

Agent operations skills — set up a project's memory layout once, rebuild working context at session start, and turn each session's lessons into approved edits to the memory structure the codebase already has

by Eric Le

Automated governance, versioning, and discovery tools for Claude Code.

by Christopher Boone

Guide for creating new plugins in this repository with consistent structure and conventions.

by Christopher Boone

Guide for creating new plugins in this repository with consistent structure and conventions.

by donald2028

A reusable skill collection for Claude Code, Codex, Cursor, Gemini CLI, Kimi Code, OpenCode, WorkBuddy, and compatible agents.

by EkLine

Open-source documentation & technical-writing skills. Vendor-neutral core — style, terminology, link checks, coverage, changelogs — plus optional EkLine power-ups.

by Allen Eubank

Language, tooling, and delivery-practice skills for agentic coding.

by Micho Calma

One development loop - FRAME, BUILD, CHECK, SHIP - replacing separate memory, planning, context, TDD and review frameworks with a single protocol, hard loop guards, and a TDD gate that is enforced by a hook rather than merely instructed.

by George Shanidze

George Shanidze's agent skills — a growing collection covering software engineering, mobile release engineering, and app store publishing. See the individual skill descriptions for what's actually in here today.

by Antonio Junior

Production-grade agent skills and custom agents for AI coding assistants.

by antonygiomarxdev

Skill plus policy directory: role definitions and gate scripts that turn one AI session into a coordinator dispatching specialised workers through Orca

by Jan Salwowski

Specification-Driven Development framework for Codex. Provides skills for feature workflows, memory system, ideas pipeline, and project scaffolding.

by Jan Salwowski

Specification-Driven Development framework for Codex. Provides skills for feature workflows, memory system, ideas pipeline, and project scaffolding.

by Francisco Perez-Sorrosal

Skills, commands, and agents for AI coding workflows — Claude Code authoring, Python development, project management, and domain-specific tools

by anortham

Miller-powered development workflow for Claude Code, Codex, and OpenCode.

by Tony Jiang

中文版:/send-to <会话名> 把消息转达给本机另一个 Claude Code 会话——会话名前缀/子串模糊匹配,多个或零个命中亮列表让用户挑,消息强制自包含,[ref] 确认照抄重发,发送后如实汇报挂起状态;跨 profile 窗口发现互不可见但 uds: 地址可直发(标准路径),配 SessionStart 注册 hook 把 socket 清单翻译成身份注册表(默认开,touch ~/.claude/.cc-session-registry-off 全局停用),四级找目标阶梯,文件交接降级为最后兜底;HAPI 会话里 ping_peer 可用但绝不传 force(被拒即 force 会杀掉对方 Claude 进程),被拒改走 inspect_peer 的 claudeSessionId → 注册表 → SendMessage;经 ping_peer 发的消息须自报家门。基于跨会话消息功能(Claude Code v2.1.224+,macOS/Linux)

by Tony Jiang

English version: /send-to <session-name> relays a message to another Claude Code session on this machine — fuzzy prefix/substring name matching, shows the session list on zero/multiple hits, enforces self-contained messages, verbatim [ref] retry, honest held-delivery reporting; cross-profile windows are mutually invisible but reachable via a direct uds: address (the standard path), a SessionStart registration hook translates the socket list into a shared identity registry (on by default; touch ~/.claude/.cc-session-registry-off to disable globally), a four-tier target-finding ladder, file handoff as the final fallback; inside HAPI sessions ping_peer may be used but never with force (a refusal means force would kill the target's Claude process) — on refusal go inspect_peer's claudeSessionId → registry → SendMessage; messages sent via ping_peer must identify the sending window. Built on cross-session messaging (Claude Code v2.1.224+, macOS/Linux)

by Aviv Sinai

Author and eval agent skills: write-skills (triggers, leading words, progressive disclosure, failure modes to prune) and eval-skills (blind golden-case runs in fresh subagents, judged against a bar, gap-driven edits). Vendored from dzhng/skills (MIT).

by Juan Cruz Fortunatti

Drive paid AI subscriptions as external agents: Codex/Grok/Antigravity/Claude Code CLIs as bounded consultants, and ChatGPT/Grok deep research through a logged-in browser. No API keys. Plus arrow-diagram: aligned ASCII flow diagrams from JSON.

by Tony Jiang

中文版:让 Claude 说人话、会提问——提问前自检清单(先核实/交代背景/黑话带解释/选项留组合位) + 表达纪律(S1~S6,含 S6 更新日志式汇报:多项进展或有未做/问题时用「本次完成/未完成/已知问题」三段体),规则源自作者 548 次真实 AskUserQuestion 抉择记录的数据挖掘。touch ~/.claude/.speak-human-always 即常驻