Agent Plugins Marketplace

Recently added

Plugins grouped by the day they were added to the directory, newest first.

Sep 19, 2026

  1. cache-warm0stars

    Keeps the 1-hour prompt cache warm for a window you set with one cache-shared fork per idle stretch, arms that for you after a paid cold write, and shows the cache state, the cold price and this session's cold writes. Needs function hooks (early access).

    Claude Code
  2. desk-companion2stars

    Live Claude Code dashboard with an animated companion, served to a phone on your Wi-Fi

    Claude Code2 skills
  3. design-playbook2stars

    Design I/O for coding agents: controllable UI generation via declarations (spec/domain/craft/design/components/template) and contracts (skill/evaluator). Use for product UI—console, dashboard, agent-ops, CJK-first apps.

    CodexClaude Code8 skills2 MCP servers
  4. human-training0stars

    Sprint-based AI-collaboration workflow skills (project genesis, lifecycle, handovers, onboarding, workflow alignment) plus session-authored skills.

    CodexClaude Code19 skills
  5. kickoff-channel0stars

    Carries this project's questions to the operator's phone through the herdr-tg hub, and carries his answers back into the agent's own turn. One Unix socket; no token, no allowlist, no Telegram in this process.

    Claude Code1 MCP server
  6. secrets-veil283stars

    Masks secret values in tool results before the model reads them: named vendor variables, value-shape patterns, and high-entropy tokens. No reveal path.

    Claude Code
  7. promote-lights283stars

    Shows CI status lights above the prompt while a promote PR (base main) is open. One REST page per minute, PR-open only. Cancelled is not-pass.

    Claude Code
  8. lesson-cards283stars

    Displays lesson hint cards when tool calls match known patterns. Additive surface during pilot; the classic --strict guard keeps its 12 block-severity denies.

    Claude Code
  9. harness-maker4stars

    Per-project AI coding harness for Claude Code, Cursor, and Codex. Bundled hm-make and hm-update skills connect native Codex plugins to a matching Python engine and project generation.

    CodexClaude Code2 skills
  10. ui-ux-pro-max0stars

    UI/UX design intelligence. Searchable local database with 84 styles, 192 palettes, 74 font pairings, 25 charts, and 22 stacks (React, Next.js, Vue, Nuxt.js, Nuxt UI, Svelte, Astro, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui, Jetpack Compose, Angular, Laravel, JavaFX, WPF, WinUI, Avalonia, Uno Platform, UWP, Three.js). Use when designing, building, or reviewing UI: pages, components, color schemes, typography, layout, accessibility, animation, or data visualization.

    Claude Code
  11. agentic-software-development0stars

    A contract-first build pipeline: plan, fan out blind concurrent agents, arbitrate, review, extract ADRs

    Claude Code1 skill
  12. other-models0stars

    Call another model harness (Gemini via Antigravity agy, OpenAI Codex) headlessly: prompt in, bare answer out.

    Claude Code1 skill
  13. likec40stars

    C4 architecture models in LikeC4 that stay true to the code: evidence rules, fact-gathering, drift maintenance and review, plus the LikeC4 MCP server for querying the model.

    Claude Code1 skill1 MCP server
  14. garden0stars

    Explicit commands for any git repo that has or is a knowledge base: init, survey, tend, maintain, review, capture, restructure, dream.

    Claude Code8 skills
  15. done0stars

    Solo project management for any complex backlog or goal: init, what-now, goals. Facts by script, judgment by model.

    Claude Code3 skills
  16. context0stars

    Context management across sessions: carryout writes a self-contained handoff prompt so a fresh Claude can resume the work.

    Claude Code1 skill
  17. base-plugin0stars

    Base plugin for your marketplace. Ships the skill-creator skill, plus a sample hook to replace with your own.

    Claude Code1 skill
  18. worktree-flow0stars

    主分支保护 + worktree 隔离边界 - main/master 默认走 worktree 或由 Human 本轮批准直写；撞闸先自救不甩锅；Codex 侧只有 skill，无机械门控

    CodexClaude Code2 skills
  19. working-discipline0stars

    AI 工作纪律 - 纪律分三层注入（SessionStart 会话级 / 每轮零章与自查 / 子代理版），下沉 4 个 guard：3 道硬拦 + 1 道事后提醒。零 skill，适合全局开启。设计依据与实测见 README

    Claude Code
  20. wenyan-output-style0stars

    文言极简输出风格 - 文言语法压缩、字形一律简体，代码与报错原样，安全告警与落盘产出物退回白话。取自 caveman 插件的 wenyan-ultra 档位，拉平成纯注入，/plugin 启停即切换

    Claude Code
  21. token-saver0stars

    运行时提醒 AI 节用 token：SessionStart 注入极简正向纪律，PreToolUse 在整读大文件时软提醒先定位行号，零硬拦。TOKEN_SAVER=off 关闭

    Claude Code
  22. task-keeper0stars

    多实例 keeper 并行托管 debug/杂务两队列：一条 bug 一个实例，原子认领 + 合并锁，登记前跨队列查重，规格空白转出互链，细节见 README

    Claude Code5 skills
  23. session-auto-title0stars

    会话标题自动跟随话题 - hook 数轮数与 assistant 行数，任一到阈值就后台调 Haiku 生成、下一轮回填 sessionTitle，补内建 ai-title 只命名一次的缺口。细节见 README

    Claude Code
  24. readable-citations0stars

    引用别处章节要免跳可读 - 每轮注入自足引用规约，让 AI 引用其它文档的章节时必带标题原文与 20 字提要、写明引用关系，链接按落点分两轨：跨平台对话正文用 file:/// 绝对路径，落盘 md 用相对路径 + 锚点。1.1.0 补落盘 md 源码出口；1.2.0 适配 Windows 终端路径。零 skill，纯注入不拦截（READABLE_CITATIONS=off 可关）

    CodexClaude Code
  25. prompt-engineering0stars

    提示词工程专家 - 优化 LLM 提示词、设计命令/钩子/技能、提升 AI 输出质量

    CodexClaude Code1 skill
  26. portable-shell0stars

    跨平台 shell 脚本 lint - Write/Edit 生成 shell 时扫 GNU(Linux)/BSD(macOS) 不可移植写法并回灌改法。PORTABLE_SHELL_LINT=off 可关

    Claude Code
  27. plain-talk-output-style0stars

    说人话输出风格 - 一语中的、几句话讲明白、能表格不成段；不专业化不掉书袋不自造词。/plugin 启停即切换风格

    Claude Code
  28. omp0stars

    Oh my pi (omp) CLI 工具集成 - 在 Claude Code/Codex 中调用 omp 实现编码、审查、搜索等任务（1个技能 + 3个子代理）

    CodexClaude Code1 skill
  29. insight-addon0stars

    教学洞察附加件 - 在真正做了技术判断的地方给一个 ★ Insight 框。只加一条规则、不定义整体风格，可与 plain-talk / adhd 等任意输出风格同时开启

    Claude Code
  30. heal0stars

    技能修复工具 - 根据执行中发现的问题更新 SKILL.md 及相关文件，支持审批工作流

    CodexClaude Code1 skill
  31. find-skills0stars

    帮助发现和安装 agent 技能 - 基于 skills.sh 生态系统，使用 npx skills CLI 搜索和安装技能。

    CodexClaude Code1 skill
  32. discover-unknowns0stars

    发现你的未知 - 与 Claude 协作挖掘未知的方法论：动手前用盲点扫描/头脑风暴/访谈/参考暴露未知；测验默认跳过、仅在用户明确要求时才出。3个 SKILL + 1个 UserPromptSubmit 未知雷达 hook（brainstorm 优先委派 superpowers:brainstorming，未安装时用内置指令）

    CodexClaude Code3 skills
  33. devkit-tool0stars

    工具技能集：代码库分析、依赖排查、大头桶核实、耗时/ROI 报告、CLAUDE.md 初始化、codegraph 建图、插件缓存刷新、submodule 拉取与嵌套推送、起会话、孤儿进程清理、订阅恢复、话题可视化（13 技能）

    CodexClaude Code13 skills
  34. devkit-spec0stars

    规范驱动开发工具集 - Spec 分析、任务拆解、实施、审查与缺陷修复（5个技能）

    CodexClaude Code5 skills
  35. content-research-writer0stars

    内容研究写作助手 - 协助研究、引用、改进钩子、迭代大纲、提供实时反馈. Includes 1 skill.

    CodexClaude Code1 skill
  36. complexity-first0stars

    复杂度优先 - 动手新增前先证明必要性：搜已有代码/标准库/平台原生，全不成立才写最小新增实现，保留安全与无障碍边界

    CodexClaude Code1 skill
  37. clickable-paths0stars

    文件路径写成可点击链接 - 注入一段输出格式规约，让 AI 提到本机文件时输出 [文件名:行号](file:///绝对路径#行号)，跨平台/终端渲染成 OSC 8 超链接直达编辑器。1.3.0 双挂子代理；1.4.0 收紧措辞圈定适用面；1.5.0 支持队列编号；1.6.0 区分落盘 md；1.7.0 适配 Windows Terminal/pwsh 盘符正斜杠与前缀路径归一化。纯注入不拦截（CLICKABLE_PATHS=off 可关）

    CodexClaude Code
  38. cd-blocker0stars

    独立 cd 拦截 - 裸 cd 开头污染会话 cwd 时硬拦并回灌可照抄改法，与写侧约束顶死的项目里可单独停用。CD_GUARD=off 单次逃生

    Claude Code
  39. agent-browser0stars

    浏览器自动化最佳实践 - 用 agent-browser（Vercel）做网页操作、数据抓取、表单填写、截图取证；默认 headless，鉴权前置 + 实例上限 4 + 安全边界。

    CodexClaude Code1 skill
  40. adhd-output-style0stars

    ADHD 友好输出风格 - 首行给可执行动作、多步编号、每轮重述进度、具体时间估算、错误陈述式语气。复刻自 ayghri/i-have-adhd（MIT），拉平成纯 SessionStart 注入，/plugin 启停即切换

    Claude Code
  41. tech-writer0stars

    Review and rewrite documentation in place using either Google's Technical Writing guidelines or strict Simplified Technical English (ASD-STE100), chosen per document

    CodexClaude Code1 skill
  42. slop0stars

    Rewrite any text to maximally overuse every known AI-writing tell — em-dashes, rule of three, "not X but Y", and more — /slop

    CodexClaude Code1 skill
  43. ship0stars

    Stage, commit, and push all changes in one step, with auto-generated message if none given

    CodexClaude Code1 skill
  44. roadmap0stars

    Generate a self-contained interactive HTML Gantt chart from a markdown roadmap file

    CodexClaude Code1 skill
  45. pr0stars

    Create a GitHub PR for the current branch with auto-generated title, summary, and test plan

    CodexClaude Code1 skill
  46. plan0stars

    Lightweight markdown-based tracker for bugs, features, chores, and todos that lives in docs/plan/ inside your repo

    CodexClaude Code5 skills
  47. okf0stars

    Author, index, and validate Open Knowledge Format (OKF v0.2) bundles — markdown concepts with YAML frontmatter for provenance, trust, lifecycle, and attested computations — so humans and agents share one vendor-neutral knowledge corpus

    CodexClaude Code4 skills
  48. monograph0stars

    Build a multi-page PhD-level GitHub Pages site on any topic — research, free-licensed photography, a distinctive topic-tuned design, and shipped via parallel expert subagents

    CodexClaude Code1 skill
  49. maintenance0stars

    Maintenance skills for the mikersays-plugins marketplace

    CodexClaude Code3 skills
  50. issues0stars

    Per-issue bug/feature/incident tracker that lives in docs/issues/ — one file per ticket with symptom, repro, root cause, fix, and verification. Includes a branch-on-start workflow and a hard rule to align with the user before implementing.

    CodexClaude Code3 skills