Engine¶

The engine is about technical means—the tools that make the paradigm work.

File System Protocol¶

Simple, flexible, powerful, stable.

The xContext Engine exposes context through a file system interface that agents already understand. Three core operations:

Browse: Navigate the context library like a file system. Directories, files, paths—familiar patterns that agents know how to use.

Search: Find what you need across the library. Different search types for different needs.

Commands: Attach capabilities to paths. Write a command to an entity to act on it. Precise, minimal, no injection problems.

Skills¶

How capabilities are organized and exposed.

Integrate: Compose skills from smaller pieces. Build up complex capabilities from simple, reliable components.

Advertise: Each skill advertises what it does. The agent discovers capabilities through honest descriptions.

xContext Skills vs. Claude Skills¶

Claude Skills have structural limitations:

• Patchwork architecture—familiar paradigms that don't connect well at boundaries
• Deep security and IP disclosure concerns

xContext Skills provide:

• Clean remote protocol—standard MCP server, no local execution
• Precise integration—commands attached to paths, no bash brittleness
• Auditable—everything that happens is visible and reviewable