AI
Saphira Linux aiDragon
A practical Saphira workspace for AI agents: your code, services, databases and tools stay under your control, while the model can run wherever it works best.
MCP: controlled tools and context
Model Context Protocol (MCP) gives an agent controlled tools and sources of context beyond the basic conversation. An MCP server can expose filesystem access, Git and project information, databases, documentation, development tools, infrastructure information and persistent memory. It is a way to make an agent useful without handing it an unstructured pile of credentials.
Saphira ships its own documentation context server for exactly this purpose. saphira-sysref 0.1.0 packages the sysref tool: a minimal local man-page context server over both the command line and MCP. Instead of dumping whole pages, an agent pulls focused man-page and Linux documentation context with context-reduction semantics, which is precisely the kind of agent-driven operating-system work Saphira itself is built with. It ships its own sysref.1 man page and a locked-down empty configuration skeleton at /etc/sysref/sysref.conf.
sysref QUERY
# Ask for focused context on one topic, not the whole manual.