135+ Built-in Tools

When you ask your AI assistant to "check the weather," "find that PDF from last month," "turn off the living room lights," or "click the save button in Photoshop", it needs actual capabilities to make those things happen. That's why Meggy ships with 135+ tools organized across 27 providers — not just text generation, but real actions that interact with your files, your network, your devices, your operating system, and the web.

Every tool invocation passes through a trust-based approval system before execution. You stay in control of what happens on your machine.

Tool Providers

Tools are organized into logical provider groups:

Provider Tools Examples
Filesystem Read, write, list, move, delete, glob, search files read_file, write_file, list_directory, glob_search
Workspace Project-level operations and code analysis workspace_summary, find_references
Terminal Execute shell commands, stream output run_command, run_background_command
Browser Headless Chromium automation, AI-driven actions, Cloudflare Browser Rendering, WebMCP bridge, accessibility snapshot navigate, screenshot, webmcp_discover, accessibility_snapshot
Desktop Computer Use native automation, hybrid perception, accessibility tree extraction, and input simulation with user-controlled application unblocking desktop_read_ui_tree, desktop_click, desktop_type
Web Search Search the web, deep multi-step research, save results to Vault web_search, deep_research
Web Fetch Retrieve and parse web page content, with SSRF protection that blocks private IP ranges and cloud metadata endpoints web_fetch, web_extract
Git Stage, commit, diff, log, branch operations git_status, git_diff, git_commit
Scheduler Create cron and one-shot scheduled tasks cron_create, cron_list, cron_toggle, set_timer
Vault Document ingestion, URL ingestion, semantic search, knowledge graph vault_search, vault_ingest, vault_ingest_url, vault_list
Memory Store and retrieve facts and context memory_upsert, memory_list, conversation_recall
Smart Devices Discover and control IoT devices on your network scan_devices, control_device, wake_on_lan
Creative Image generation via multiple providers generate_image
Presentations Export slide decks to PPTX/PDF presentation_export
Network DNS lookup, HTTP requests, port scanning dns_lookup, http_request
System Monitor CPU, memory, disk, and process monitoring system_info, process_list
Outbound Messaging Send messages across connected channels send_message, send_channel_message
People Manage the People Knowledge Graph people_search, people_set_role, people_add_relationship
Converter Unit and format conversion utilities convert_units, convert_currency
Date & Time Timezone, calendar, and scheduling utilities current_time, time_difference
Generator Random data, UUIDs, passwords generate_uuid, random_password
Questions Cross-channel interactive question engine ask_user, ask_channel
Admin Administrative and configuration operations get_settings, update_config
Weather Current conditions and forecasts get_weather, weather_forecast
Clipboard Read and write system clipboard clipboard_read, clipboard_write

The Approval System

Having 130+ tools is powerful — but you don't want your AI deleting files without asking. That's why Meggy implements a three-tier trust model:

Trust Tiers

Tier Who Permissions
Owner The primary user Full access — can approve dangerous tools
Trusted Vetted agents and channels Pre-approved for safe/moderate tools
Unknown New or unverified agents Requires explicit approval for every tool call

Risk Classification

Every tool is classified into one of three risk levels:

Approval Modes

Meggy supports multiple approval modes:

MCP Integration

Beyond built-in tools, Meggy supports loading external tools via the Model Context Protocol (MCP). MCP servers expose tools, resources, and prompts that the AI can discover and invoke — extending Meggy's capabilities without modifying its core. This means the tool ecosystem grows with you.

What's Next?