# Cap learning extraction to prevent infinite loops

> Learning stores add an extraction_tool_call_limit that bounds how many tool calls the extraction step can make.

- Published: 2026-07-23
- Author: Agno Team
- Category: Changelog
- Canonical: https://agno-com-nine.vercel.app/articles/cap-learning-extraction-to-prevent-infinite-loops
- Markdown: https://agno-com-nine.vercel.app/articles/cap-learning-extraction-to-prevent-infinite-loops.md

Learning stores add an `extraction_tool_call_limit` that bounds how many tool calls the extraction step can make. The extraction process stops at a sensible ceiling rather than spinning indefinitely, so a runaway loop can't rack up cost or stall a run.