NVIDIA/NeMo-Agent-Toolkit v1.6.0

v1.6.0

Repository: NVIDIA/NeMo-Agent-Toolkit

Tag: v1.6.0

Published: 2026-04-10T02:27:16Z

Prerelease: no

Release notes:

🚀 Notable Features and Improvements

• This release introduces ATIF-based evaluation, including a standalone evaluator class that can evaluate ATIF trajectories generated by any system

What's Changed

🚨 Breaking Changes

• NAT Eval Dependency Reduction by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1690
• Migrate nvidia_nat_ragas to Ragas v0.4 by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1780
• Add support for optionally disabling SSL/TLS verification by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1640
• Add centralized user identity management and custom WebSocket authentication by @ericevans-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1775
• Feat/optimizer package extraction by @bhuvan002 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1637
• Decouple ATIF schemas from core and make nvidia-nat-eval lightweight by default by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1833

✨ New Features

• Add TimeoutMiddleware by @ericevans-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1677
• feat(agent): stream tool call chunks by @MylesShannon in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1717
• Add ATIF adapter: IntermediateStep to Harbor ATIF v1.6 trajectory conversion by @yczhang-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1663
• feat(control_flow): add built-in parallel_executor with example by @antoniomtz in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1733
• Restore TimeoutMiddleware reverted by #1710 by @ericevans-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1753
• Add ATIF-native evaluator interface and runtime dual-lane dispatch by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1747
• Add LLM error recovery, retry logic, and diagnostic logging to Tool Calling Agent by @ericevans-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1757
• Add ATIF-only custom evaluator example and docs by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1766
• feat(profiler): ATIF-native profiler by @afourniernv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1751
• Add pre-tool verifier defense middleware for input instruction violation detection by @lidan-capsule in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1605
• Add MemMachine memory integration for NeMo Agent toolkit by @Charlie-Yi-2002 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1460
• Nested ATIF lineage by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1814

🔧 Improvements

• Forward-merge release/1.5 into develop by @willkill07 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1676
• Add user attribution to Weave traces and feedback by @thepatrickchin in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1668
• Move eval callbacks to nvidia_nat_eval and guard core shim path by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1748
• Refactor eval file output into FileEvalCallback for Python API support by @yczhang-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1743
• Add AtifBaseEvaluator for reusable concurrent ATIF custom evaluators by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1769
• Use nvidia-nat-eval from PyPI in the custom evaluator notebook by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1778
• chore(pre-commit): any pyproject.toml changes should trigger uv lock checks by @willkill07 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1790
• Remove unneeded warning filter by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1789
• Remove old/unused compatibility code by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1794
• fix: handle Anthropic-style tool schemas in LangchainProfilerHandler by @cluster2600 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1770
• Add ATIF-native evaluators for nvidia-nat-profiler runtime metrics by @afourniernv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1791
• Update old CI variable names by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1813
• Explicitly set framework_wrappers for LangChain tools by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1820
• Report installed packages when testing built wheels by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1821
• chore: pin LiteLLM to known good version by @willkill07 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1823
• Ensure file paths are properly quoted in gitutils.py by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1832
• Update LICENSE-3rd-party.txt by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1840
• Informational ATIF mapping docs and sample output colocation (no functional changes) by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1844
• fix: guard nim_langchain against ChatNVIDIA model lookup crash by @bbednarski9 in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1843
• Document that MemMachine integration is Experimental by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1854
• Restore profiler intermediate-step compatibility by rolling back selected ATIF-native changes by @AnuradhaKaruppiah in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1853

🐛 Bug Fixes

• Fix Simple Calculator MCP example to return consistent timezone results by @dagardner-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1708
• fix: preserve nested Pydantic models when unpacking multi-arg functions by @MylesShannon in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1696
• Fix _running_workflow_task always being None due to method chaining by @AjayThorve in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1744
• Fix LoRA inference routing in OpenPipe ART cookbook by @aslanshi in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1662
• Add profiler callback to sequential executor for LLM observability by @thepatrickchin in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1786
• Enhance chat completion functionality to support OpenAI-style message history by @mdemoret-nv in https://github.com/NVIDIA/NeMo-Agent-Toolkit/pull/1674

*…