Forward-merge release/1.4 into develop

[willkill07]

Description

Closes #1545

By Submitting this PR I confirm:

• I am familiar with the Contributing Guidelines.
• We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license.
  • Any contribution which contains commits that are not Signed-Off will not be accepted.
• When the PR is ready for review, new or existing tests cover these changes.
• When the PR is ready for review, the documentation is up to date with these changes.

Summary by CodeRabbit

• Documentation
  • Published Release 1.4.0 notes with comprehensive feature and improvement documentation
  • Updated product positioning to emphasize AI agent intelligence, observability, and continuous learning
  • Reorganized documentation structure to highlight agent frameworks, optimization capabilities, and security features
  • Updated roadmap with KV-Cache optimization and memory enhancements

[coderabbitai]

Walkthrough

Forward-merge of release/1.4 branch into develop, updating documentation for the 1.4.0 release. Changes include new changelog entries, restructured README with updated feature descriptions and roadmap, and refined titles across multiple documentation guides. No code or API changes.

Changes

Cohort / File(s)	Summary
Release Documentation CHANGELOG.md, docs/source/release-notes.md	Added comprehensive v1.4.0 release notes and restructured release documentation with new feature descriptions (LangGraph wrapper, RL improvements, Dynamo integration, security engine, A2A support, Bedrock/AutoGen support).
Project Overview README.md	Updated tagline and reorganized feature sections into "Building Agents" and "Agent Insights/Optimization" subsections; revised roadmap with KV-cache optimization and memory interface improvements; adjusted installation documentation wording.
Documentation Guide Updates docs/source/improve-workflows/evaluate.md, docs/source/improve-workflows/finetuning/index.md, docs/source/improve-workflows/finetuning/rl_with_openpipe.md, docs/source/improve-workflows/optimizer.md	Updated guide titles for clarity and consistency (e.g., "Agent Evaluation", "GRPO with OpenPipe ART", "NVIDIA NeMo Agent Toolkit Finetuning Harness for Reinforcement Learning"); minor formatting adjustments in markdown structure.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~15 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'Forward-merge release/1.4 into develop' is fully related to the changeset, clearly summarizing the main change of merging release/1.4 into develop while meeting the requirements for conciseness and imperative mood.
Linked Issues check	✅ Passed	The PR successfully addresses issue #1545 by forwarding changes from release/1.4 into develop, keeping the branches synchronized as required by the linked issue.
Out of Scope Changes check	✅ Passed	All changes in the PR are in-scope forward-merge updates (CHANGELOG.md, README.md, and documentation files) that reflect release 1.4.0 content synchronized from the release branch.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[mnajafian-nv]

LGTM

[coderabbitai]

Actionable comments posted: 5

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

packages/nvidia_nat_langchain/tests/test_langchain_agents.py (1)

26-27: ⚠️ Potential issue | 🟡 Minor

Remove the global HuggingFaceConfig import to avoid redefinition and unguarded optional deps.

The local import is correctly guarded by the optional-dependency check, but the top-level import now triggers Ruff F811 and defeats the purpose of lazy imports. Drop the module-level import.

🔧 Proposed fix

-from nat.llm.huggingface_llm import HuggingFaceConfig

Also applies to: 157-157

🤖 Fix all issues with AI agents

In `@CHANGELOG.md`:
- Around line 54-60: Update the CHANGELOG bullets that use the acronym "NAT" so
they either spell out "NeMo Agent Toolkit" or wrap code identifiers in
backticks; specifically replace "NAT A2A Client & Server Support" (and any other
bullets containing "NAT" such as the bridge/implementation bullets) with "NeMo
Agent Toolkit A2A Client & Server Support" or use backticks for code identifiers
like `NAT` and ensure "a2a" references are consistently formatted as either
spelled out or backticked (e.g., `A2A`) to comply with the Markdown
documentation guideline.

In `@docs/source/improve-workflows/optimizer.md`:
- Around line 344-350: Remove the leftover Git conflict markers (<<<<<<<,
=======, >>>>>>>) in the two prompt entries and keep the intended module path
`~nat.agent.prompt_optimizer.register`; specifically update the
`prompt.prompt_population_init_function` and
`prompt.prompt_recombination_function` lines to a single, clean pair of bullets
using `~nat.agent.prompt_optimizer.register` (no conflict markers), and ensure
the text follows the doc style guidelines (use "NeMo Agent Toolkit" for
titles/pages and "NeMo Agent Toolkit" or "the toolkit" elsewhere as
appropriate).

In `@README.md`:
- Around line 68-71: The README uses an inanimate possessive in the "Agent
Optimization" section ("Improve your agent's quality, accuracy, and
performance"); update that sentence to remove the "'s" (e.g., change to "Improve
agent quality, accuracy, and performance" or "Improve the quality, accuracy, and
performance of your agent") so the header/intro line in the "Agent Optimization"
block is non-possessive while leaving the linked bullet items (Evaluation
System, Hyper-Parameter and Prompt Optimizer, Fine-tuning with Reinforcement
Learning) unchanged.
- Around line 44-45: Change the inanimate possessive "agent’s workflows" in the
README sentence under "[**Automatic Reinforcement Learning (RL):**]" to a
non-possessive form such as "agent workflows" or "workflows of the agent"; edit
the phrase inside the sentence that currently reads "fine-tuning open LLMs to
better understand your agent's workflows, tools, and prompts" so it reads e.g.
"fine-tuning open LLMs to better understand your agent workflows, tools, and
prompts" (or use "workflows of the agent") to comply with the Markdown
documentation guideline.
- Around line 61-63: Update the sentence in the "Framework Agnostic" bullet that
currently reads "Use the toolkit with popular frameworks..." to use a
capitalized product name—replace it with "Use the Toolkit with popular
frameworks..." or "Use NeMo Agent Toolkit with popular frameworks..."; locate
the exact string "Use the toolkit with popular frameworks such as ..." in the
README.md and make the capitalization change there so documentation uses
"Toolkit" consistently.

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@docs/source/release-notes.md`:
- Line 26: The phrase "your agent's workflows, tools, and prompts" in the
release notes uses an inanimate possessive; update that sentence to avoid "'s"
by replacing that fragment with "agent workflows, tools, and prompts" (or
"workflows of your agent, tools, and prompts") so the line reads with the
non-possessive form while preserving the rest of the sentence and links.

[willkill07]

/merge nosquash

[rapids-bot]

Could not determine original ForwardMerger PR from branch name. The branch name should follow the pattern <target_branch>-merge-<source_branch> (e.g., branch-25.06-merge-branch-25.04 or main-merge-release/25.06). Please contact @rapidsdevops on Slack for assistance.