# Product concepts

> Mental models for Switchyard projects, runtimes, operations, trust, evidence, agents, workspaces, and adapters.

Switchyard models a repository and its local operational environment as a
project. The [glossary](../glossary.md) defines the stable vocabulary for
projects, services, runtimes, runs, operations, actions, declarations,
reservations, and bindings.

The [architecture overview](../architecture/README.md) maps those product
concepts to bounded domains and adapters. Start with the focused concept that
matches the question:

- [Project onboarding and explicit trust](../architecture/project-onboarding.md)
- [Durable operations](../architecture/operations-kernel.md)
- [Health, logs, resources, and retained evidence](../architecture/observability.md)
- [Ports, Git, and trusted actions](../architecture/developer-workflows.md)
- [Coding-agent permissions and MCP](../architecture/agent-integration.md)
- [Workspaces and worktrees](../architecture/workspaces.md)