Plugins

You just set up Meggy and want it to help with your kids' homework. You could dig through the skills list, find the right ones, enable them one by one, configure content safety, and set up hooks for notifications. Or you could flip one switch.

Meggy's Plugin system bundles everything you need for a specific use case — skills, tools, safety filters, hooks, agents, even vault folders — into a single package. Enable the Kids Mode plugin and you instantly get 4 education skills, strict content filtering, and age-appropriate language guidelines. Disable it when the kids go to bed and everything reverts.

Think of plugins as feature packs — pre-configured combinations that turn your general assistant into a domain specialist in one click.

What's Inside a Plugin?

Each plugin can bundle any combination of:

Component What It Does
Skills Domain knowledge that shapes how the AI responds
Tools Access to specific capabilities like terminal, smart devices, or file operations
Hooks Automated rules that fire on events (notifications, safety gates, workflow chains)
Agents Pre-built AI agents configured for specific tasks
Safety settings Content filter levels that protect sensitive audiences
Vault folders Organized storage spaces for the plugin's domain

When you enable a plugin, all its components activate together. When you disable it, everything deregisters cleanly — no leftover configuration, no orphaned hooks.

Built-in Plugins

Meggy ships with 14 built-in plugins spanning 8 categories:

Plugin Category What You Get
📚 Teaching & Tutoring Education 4 skills for Socratic tutoring, exercise generation, vocabulary building + strict content filter
👶 Kids Mode Education 4 skills with age-appropriate language + strict content filter + behavioral guidelines
👴 Elder Care Home 2 skills for medication reminders, simplified instructions, and patient communication
🏠 Home Automation Home 3 skills + smart device hooks for lights, plugs, and sensors
🧹 Home Management Home 4 skills for chore rotations, cleaning schedules, and household organization
🔨 Home Renovation Home 3 skills for project planning, contractor coordination, and material estimation
🏥 Doctor & Health Health 3 skills + medical disclaimer (always reminds users to consult a professional)
💪 Fitness & Training Health 3 skills for workout planning, progress tracking, and form guidance
💼 Work Productivity Productivity 11 skills + Google Workspace integration for email, calendar, and docs
💻 Developer Toolkit Coding 17 skills + terminal access with human-approval safety gate
💰 Finance & Budgeting Finance 4 skills for budget tracking, savings goals, and financial planning
✍️ Content Creator Writing 7 skills for copywriting, editing, social media, and content strategy
📊 Sales & Research Business 4 skills for market analysis, prospect research, and competitive intelligence
🖥️ System Administration DevOps 5 skills + terminal access with approval gates for server management

All 14 come pre-installed and disabled by default. Enable what you need, ignore what you don't.

Enabling a Plugin

  1. Open Settings → Plugins
  2. Find the plugin you want
  3. Toggle it on
  4. That's it — all skills, tools, hooks, and safety settings activate instantly

Disabling works the same way. One toggle off and every component deregisters.

You can also manage plugins from any chat channel — Desktop, Telegram, WhatsApp, or voice. Just say list my plugins, enable the developer plugin, or disable the kids mode plugin. Meggy handles everything conversationally.

Importing Plugins from GitHub

Beyond the 14 built-in plugins, you can import plugins directly from GitHub URLs — including the Anthropic Cowork Knowledge Work Plugins (Productivity, Sales, Data Analyst, Marketing, and more).

Just ask Meggy:

import the plugin from https://github.com/anthropics/knowledge-work-plugins/tree/main/productivity

Meggy will:

  1. Preview the plugin — show you what skills, commands, hooks, and MCP servers it includes
  2. Ask for confirmation before installing
  3. Import all components automatically
  4. Prompt for credentials if any MCP servers need API keys (e.g., database connectors)
  5. Enable the plugin so its skills and tools are immediately available

This works from any chat channel — you don't need to be at the desktop app. Paste a GitHub URL in Telegram and Meggy handles the rest.

Content Safety

Some plugins — like Kids Mode and Teaching & Tutoring — include a content safety filter. When you enable them, Meggy raises its content filter level automatically.

The key design principle: safety can only go up, never down. If you enable Kids Mode (strict filter) alongside Developer Toolkit, you get coding assistance with family-safe language. Disable Kids Mode and the filter recalculates based on whatever plugins remain active.

This means you never accidentally expose a child to inappropriate content just because another plugin is also enabled.

How Meggy Plugins Compare to Claude's

If you've used Anthropic's Claude, you might be familiar with their plugin model. Here's how the two approaches differ:

Feature Meggy Plugins Claude Plugins
What they bundle Skills + tools + hooks + agents + safety + vault folders Slash commands, subagents, MCP servers
Where they run Fully local — nothing leaves your machine Cloud-hosted with MCP server connections
Privacy Your data stays on your device Data passes through Anthropic's servers
Content safety Built-in safety filters per plugin (escalation-only) Relies on API-level content policies
Plugin source 14 bundled + Claude import + manual install Official marketplace + community repos
Hook automation Event-driven hooks bundled with plugins Hooks via MCP Apps and custom scripts
Offline support Works fully offline with local models Requires internet connection

Claude Plugin Import

Already have plugins from Claude? Meggy can import them. The Claude Plugin Importer reads Claude-compatible plugin directories and converts their hooks, MCP servers, commands, and skills into Meggy's native format.

Claude's hook events map directly:

Claude Event Meggy Event
PreToolUse preToolCall
PostToolUse postToolResult
PreAssistantMessage beforeModel
PostAssistantMessage afterModel

Components that don't have a direct Meggy equivalent are flagged as warnings during import so you know what needs manual attention.

This importer works for both local directories and GitHub-hosted plugins — including the full Anthropic Cowork plugin collection (productivity, sales, data analysis, customer support, marketing, legal, finance, and more).

Plugin Health

Every enabled plugin gets a health dashboard showing:

If a plugin has conflicts with another enabled plugin — say two plugins both try to configure the same tool — Meggy flags it and tells you which one takes priority.

What's Next?