All notable changes to AI Maestro Plugins are documented in this file. Format follows Keep a Changelog.
- Modular CLI architecture -- Split
aimaestro-agent.sh(3,879 lines) into 6 focused modules:agent-core.sh-- Shared infrastructure: security scanning, validation, JSON editing, Claude CLI helpersagent-commands.sh-- CRUD: list, show, create, delete, update, rename, export, importagent-session.sh-- Session lifecycle: add, remove, exec, hibernate, wake, restartagent-skill.sh-- Skill management: list, add, remove, install, uninstallagent-plugin.sh-- Plugin management (10 subcommands) + marketplace (4 subcommands)aimaestro-agent.sh-- Thin 108-line dispatcher that sources all modules
- Each module has a double-source guard and uses the fallback path pattern (
SCRIPT_DIRthen~/.local/bin/) - No functional changes -- all commands work identically
- Initial plugin release with 6 skills and CLI scripts
- Agent lifecycle management (create, delete, hibernate, wake, rename, export, import)
- Plugin and marketplace management (install, uninstall, update, enable, disable)
- Skill management (list, add, remove, install, uninstall)
- AMP messaging integration
- Code graph querying
- Memory search
- Documentation search
- Planning skill with persistent task tracking
- ToxicSkills security scanner for skill installation
- Auto-trust mechanism for agent creation
AIM_AGENT_*environment variables (replacingCLAUDE_AGENT_*with backward compat)