server_stats
Gather OS-level server statistics for system prompt injection.
All heavy reads (/proc, psutil) are wrapped in
asyncio.to_thread() so the event loop is never blocked.
Results are cached for 30 seconds to avoid redundant psutil
reads on every inference.
-
async server_stats.get_server_stats(background_task_count=0)[source]
Return a dict of live server statistics.
Results are cached for 30 seconds to avoid redundant psutil reads.
- Parameters:
background_task_count (int) – Number of currently active background tool tasks, provided
by the caller (typically from TaskManager).
- Return type:
dict[str, Any]