Agent Safehouse

kept by eddie mar 10

Agent Safehouse is a macOS-native sandbox tool that uses kernel-level enforcement to deny LLM coding agents access to all files except explicitly granted directories.

Agent Safehouse – macOS-native sandboxing for local agents
Agent Safehouse wraps macOS sandbox-exec to sandbox local AI agents with presets, enabling secure full-auto operation.
also on Agent Safehouse, #macos, #cli, #security, #ai-agents, #developer-tools
Homebrew: 6.0.0
Homebrew 6.0.0 ships tap-trust security, default internal JSON API, Linux sandboxing, bundle parallel installs, and deprecates Intel macOS for 2027.
also on Homebrew, #macos, #security, #developer-tools
GitHub - googleworkspace/cli: Google Workspace CLI — one command-line tool for Drive, Gmail, Calendar, Sheets, Docs, Chat, Admin, and more. Dynamically built from Google Discovery Service. Includes AI agent skills.
Google releases a dynamic CLI for Workspace APIs that builds commands from the Discovery Service at runtime and includes 100+ AI agent skills.
also on Homebrew, Gemini, #cli, #ai-agents, #developer-tools
A native macOS terminal app for AI agents.
Unpeel is a native macOS terminal app that orchestrates multiple AI agent CLIs side by side with built-in MCP servers, persistent sessions, and remote access.
also on #macos, Gemini, Codex, #cli, Claude, #ai-agents, #developer-tools