VibesPlug skill
A bundle of 176 skills, subagents and an MCP server for Claude Code, Cursor, Codex and Windsurf with code checks
- 121 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The one-line installers run a script from the network, and the package writes rules and git hooks into project and editor directories
- The MCP server and sandbox execute code, and the REA set analyzes binaries through Ghidra and LLDB
- The payments and CI skills let the agent change sensitive code and configuration
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/vibesplugDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The one-line installers run a script from the network, and the package writes rules and git hooks into project and editor directories
- The MCP server and sandbox execute code, and the REA set analyzes binaries through Ghidra and LLDB
- The payments and CI skills let the agent change sensitive code and configuration
This is third-party code. Review the repository files before installing.
What it does
VibesPlug installs 176 domain skills, a set of specialist subagents in `agents/` and rules for different editors into your environment. The skills fall into groups: frontend and mobile UI, backend and API, AI and MCP, SaaS and payments, testing and security, data and DevOps. The subagents take separate roles: architect, full-stack developer, quality gate, payments, MCP, cloud and SRE, data and RAG, mobile and reverse engineering. The `vibes` CLI adds commands: `vibes mcp` starts an MCP server over stdio and SSE with dozens of tools, `vibes studio` opens a local dashboard at `http://localhost:3888`, `vibes hooks install` installs a pre-commit hook with checks, `vibes sandbox` runs a code snippet in an isolated environment and `vibes ci:heal` tries to repair a failing CI. A separate REA skill set analyzes binaries, firmware and Electron builds through Ghidra and LLDB. The set's core rule, the anti-slop directive, requires the agent to remove stubs, filler text and templated interfaces. The installer configures Antigravity, Claude Code, Cursor, Codex CLI, Windsurf, Trae and VS Code extensions.
Who it is for. Developers who run projects across several AI editors at once and want one shared set of rules, skills and checks instead of scattered prompts.
Good fit when
- Cursor, Claude Code and Codex work on the same repository and need identical project rules
- You want an MCP server with AST queries, SQL schema checks and scoring of competing solutions
- A hook should fire before each commit and reject stubs and boilerplate
- The task is reverse engineering a binary or an Electron app with checkable evidence
Not a fit when
- You want one skill for one task: the bundle is large and adds rules and hooks across the whole project
- You need to protect the repository from extra changes: the installer edits editor configs and installs git hooks
- Running scripts from the network is not allowed: the main one-line installers use `curl | bash` and `irm | iex`
Example request
Run vibes verify-gate on this file and show which stubs and security problems were foundLimitations
The savings figures in the README (75–90 percent cost reduction, 93 percent context compression, 2 ms execution) come from the author without separate measurements, so verify them on your own tasks. The bundle is large: 176 skills and dozens of MCP tools increase context size, and tool profiles (`vibes_tool_profile`) cut it by up to 75 percent by the author's claim. The installer writes rules into `.cursor`, `.claude`, `.codex` and other directories and can install git hooks. The README is heavy with promotional wording. The description and some texts come in several languages, including Indonesian, Chinese, German and Japanese.
How to disable. Remove a single skill with `vibes remove <skill-name>`, drop hooks through `vibes_git_hooks` or by deleting the file in `.git/hooks`. Remove the whole package with `npm uninstall -g vibes-plug` and by deleting the added rule directories.
FAQ
How do I install VibesPlug in Claude Code?
Run `npm i -g vibes-plug`, then `vibes-plug init` in the project. From the repository you can also run `node scripts/install.js --claude --project /path/to/my-project`.
Which editors and agents are supported?
The README names Google Antigravity, Claude Code, Cursor, OpenAI Codex CLI, Windsurf, Trae, VS Code with Cline and Roo, Kimi and Kiro. For Windsurf, Trae and Cline it is enough to copy the ready rule files into the project.
How do I connect the MCP server?
Run `vibes mcp` or `npm run mcp` and add the command to the `mcp_config.json` of Claude Desktop, Cursor, Antigravity or Codex. The server works over stdio and SSE.
How do I update installed skills?
If the repository is cloned, go to its folder and run `git pull origin main` and `npm test`. A global package is updated through npm.
How do I remove a single skill?
`vibes remove <skill-name>` removes an installed skill from the project. The git hook is installed with `vibes hooks install` and removed by deleting its file from `.git/hooks`.
Discussion
No questions or reviews yet.
Related
SkillsA kit of 138 agents, 296 skills and 74 hooks that turns Claude Code into a full software development team
WebUI and CLI daemon for remote work with local agents over ACP: Claude Code, OpenCode, Gemini CLI and 20+ more.
Vibe Check MCP
vibe-check-mcp-server
A mentor tool for an agent: it asks the agent to stop and check its plan before acting. The README states maintenance mode, and model keys are needed
AI Builder Club Skills
AI Builder Club — Skills
A Claude Code plugin marketplace to make a repo agent-ready and run compounding agent loops
One config for Claude Code and Codex in every repo: shared rules, skill packs, second-model review and a destructive-command guard
Vibe Coding Toolkit
vibe-coding-toolkit
A set of AI development techniques: plugins, subagents and wave orchestration. The README is in Portuguese
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| VibesPlug (this card) | no | 121 | |
| Agent Swarm operator skill | no | 1,012 | organizes a swarm of agents around a task without a bundle of 176 skills or an MCP server |
| OpenCode Swarm | no | 518 | a plugin for OpenCode with agent roles |
README in short
The README opens with a promotional headline, a comparison table against plain prompts and IDE rules, skill categories and a table of subagents. It then covers the P1–P8 engines, the reverse-engineering section, the MCP tool list, the Studio dashboard and install instructions in English and four more languages.
README badge
Original SKILL.md
--- name: anti-slop description: "Zero-tolerance Anti-AI Slop engine: crisp UI, clean code, no slop / Engine Anti-AI Slop: UI presisi, kode bersih tanpa placeholder." author: "Roedy Rustam" version: "4.4.4" --- # Sovereign Anti-AI Slop & Craftsmanship Engine (2026 Unified Edition) [English](#english) | [Bahasa Indonesia](#bahasa-indonesia) --- <a name="english"></a> ## English ### Executive Directive & Trigger Conditions The **Sovereign Anti-Slop Directive** is the uncompromising zero-tolerance standard governing all AI agent actions across the vibes-plug swarm. AI slop degrades developer trust, burns context tokens, bloats codebases, and introduces catastrophic production defects. Anti-slop is a **filter, not a style guide**: it prescribes no fixed colors, fonts, or layouts—your creative direction remains yours. Removing slop clears away generic AI habits, while the **Liveliness Toolkit** ensures the final product is vivid, authentic, and deliberate rather than sterile. **Universal Trigger Conditions:** - **Every Code Generation & Modification**: Every function, component, or file created or modified by any agent. - **Frontend & UI Work**: Design systems, page layouts, components, cards, dashboards, and responsive styling. - **Copywriting & Prose**: Headlines, CTAs, product copy, documentation, commit messages, and PR summaries. - **Code Comments**: Documentation comments, inline annotations, and docstrings. - **Automated Quality Gates & PR Audits**: Code reviews (`coderabbit`), TDD loops (`autonomous-tdd-debugger`), and production release hardening (`production-ready-hardener`). --- ## Two Usage Modes Protocol On every anti-slop activation, resolve the operational mode in this strict priority: 1. **Explicit Session Instruction**: User specifies `during` or `after` in the prompt. Announce once: `antislop active: <mode> (session override).` 2. **Global Preference**: Read `%APPDATA%\antislop\settings.json` (Windows) or `~/.config/antislop/settings.json` (macOS/Linux). 3. **Default Fallback**: Mode 1 (DURING) for new features/code generation; Mode 2 (AFTER) when auditing existing projects. ### Mode 1: DURING (Active Real-Time Prevention) - Follow all 38 rules actively while generating code, copy, and UI. - Prevents slop from being written in the first place. - Conclude the task with the **Mandatory Delivery Gate** report. ### Mode 2: AFTER (Post-Build Audit & Remediation) - Audit an existing or completed codebase/project without altering files prematurely. - Generate a timestamped audit report in `anti-slop/audit-001-YYYY-MM-DD.md` (incrementing sequence).