Infrastructure for agent coordination. Track agent execution, persist context across sessions, and build multi-agent workflows.
The ai-assisted-development plugin observes agent execution via hooks and stores:
- User prompts and agent outputs
- Agent-created artifacts
- Session logs pruned to request boundaries
- Complete audit trail
Agents coordinate without coupling. They output tags; the system extracts them. No special tools required.
See ARCHITECTURE.md for design details.
git clone https://github.com/macgregor/claude-code-toolbox.git
cd claude-code-toolbox
make install-pluginNavigate to any repository and use Claude Code normally. The plugin tracks execution automatically:
cd ~/your-project
claudeCheck .toolbox/events/ for captured context and artifacts.
Using:
- ARCHITECTURE.md - System design and lifecycle tracking
- docs/devcontainer.md - Isolated development environment
Building:
- CONTRIBUTING.md - Local development setup and testing
- docs/claude-code-reference.md - Platform features and constraints
Learning:
- docs/ai-learning-material.md - AI development resources