Agno 3.0.1 caches tool schemas across runs and loads session history incrementally, so response time stops scaling with conversation length
Agno 3.0.1 landed 26 August, two days after the 3.0.0 stable release, with two performance changes aimed at long-running agents: tool schemas are now derived once and cached across runs rather than rebuilt per run, and session history is loaded incrementally per turn so response time stays flat as a conversation grows instead of scaling with its length. The fix list is mostly correctness for MCP and Gemini users, including Function.process_schema_for_strict no longer raising KeyError on schemas that omit properties (which MCP server schemas registered verbatim commonly do), and tool-returned images and document blobs now nested inside FunctionResponse.parts for Gemini 3 and later rather than emitted as sibling inline_data parts. A core-only install bug where add_location_to_context=True raised ModuleNotFoundError on an undeclared requests dependency is also fixed.
↳ Follow the thread