fix: keep a completed Deep Agents turn completed when telemetry fails

[SandyChapman]

Overview

DeepAgentsRuntime.invoke() wrapped the Relay plugin/scope context managers and the agent call in a single try/except Exception. A context manager's __exit__ runs inside that try, so an exception raised purely while closing the telemetry scope was assigned to the invocation error, and normalize_output then reported completed=false / failed=true for a turn the agent had already finished — final response, tool calls, and workspace changes all intact.

This is reachable today. NeMo Relay keeps one process-global LIFO scope stack in a ContextVar; LangGraph schedules child tasks with copy_context(), which preserves that same mutable stack rather than cloning it, so concurrent chain callbacks can close out of LIFO order and Relay's scope validator rejects the pop with:

RuntimeError: invalid argument: scope handle is not at the top of the stack

The stranded child scope then makes the outer deepagents-request scope raise on exit. Downstream this scored 31 otherwise-successful NeMo Evaluator agent-eval trials as adapter invocation failures (NVBug 6562846), including at evaluator parallelism 1. Deep Agents is where this is reachable because it is the only adapter that installs a LangChain callback handler, and its nested per-chain scopes are what LangGraph's concurrent tasks close out of order.

This PR separates the two failure domains structurally rather than papering over the Relay bug:

• TurnOutcome carries error (the agent failed, no functional result) and telemetry_error (recording the turn failed) as independent fields.
• _invoke_agent owns the invocation domain. _invoke_with_telemetry owns the telemetry one and guards nothing but the context managers, so an exception it catches can only be a telemetry fault.
• Whether the agent got to run is answered by whether an outcome exists, not by a side flag. No outcome means telemetry failed on the way in and there is nothing to preserve, so that stays an invocation failure — the split cannot widen into general failure suppression.
• Artifact collection is covered too. It walks the filesystem after the turn is over, so letting it raise would discard a completed invocation — the same failure mode this change exists to stop. _telemetry_output returns that fault instead of raising, and faults from multiple stages are joined rather than the first one winning.
• telemetry.degraded: true accompanies telemetry.error, so a consumer has a machine-readable signal instead of a message to parse. Artifacts are still referenced when degraded — a partial trajectory is usually worth reading — but the flag marks them as possibly truncated. Both keys are absent on a clean run, and the block is emitted for a fault even without a runtime block so the diagnostic cannot be silently dropped.

The Relay-side root cause (shared scope stack across concurrent LangGraph tasks; a failed scope.pop stranding an untracked scope) is being reported separately against NeMo-Relay. This change is worth making on its own regardless: an observability fault should never rewrite the functional outcome of a completed invocation.

Where should the reviewer start?

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py — TurnOutcome and the _invoke_with_telemetry / _invoke_agent split. The design decision to check is the outcome is None test in _invoke_with_telemetry: it is what distinguishes "telemetry broke on the way out, after a completed turn" (recoverable) from "telemetry broke on the way in, so the agent never ran" (a real failure).

The contract question worth a maintainer's opinion is the artifact policy: on a degraded run this references artifacts anyway and flags them, rather than withholding them. Withholding is the other defensible choice; happy to switch if you'd rather a degraded run expose nothing.

Then tests/adapters/test_deepagents.py, which pins each branch: scope-teardown fault keeps the turn completed; plugin-flush fault likewise; artifact-collection fault likewise; two faults in one turn are both reported; a setup fault stays an invocation failure; a genuine agent failure still fails with a clean telemetry block; and agent failure plus teardown fault keep their own domains.

Review follow-ups from this round: the completed-result claim in the README was overstated (a telemetry fault after an agent failure correctly leaves the turn failed) and is now qualified; the setup-failure test asserts invoke_compiled_agent was never awaited, so a regression that consumes an agent request before reporting the fault is caught.

Not applied: redacting the exception text in telemetry.error. The concern is real, but this adapter already returns raw str(exc) in the invocation error and has since before this PR, and that path is the likelier credential carrier (a provider client error can echo a base_url). Redacting only the new field would be inconsistent and would remove the diagnostic the field exists for. If the exception text should be sanitized, it belongs in one place across the normalized result — happy to open that separately.

Testing:

• just test-python — 680 passed, 53 skipped.
• tests/adapters/test_deepagents.py — 49 passed (42 before, 7 new).
• Mutation checks, each caught by a test: forcing the outer handler to always keep the outcome (drops the setup-failure branch); letting artifact collection raise; reporting only the first of two faults.
• ruff check / ruff format --check (v0.15.21, matching .pre-commit-config.yaml) and the copyright-header hook pass on all changed files.

Not addressed here, flagged rather than fixed: _invoke_agent collapses an exception to f"{type}: {exc}" and discards the traceback, which is pre-existing behavior and this module has no logger to route it to. Separately, tests/adapters/test_claude_adapter.py::test_build_options_maps_normalized_capabilities_and_claude_settings fails whenever ANTHROPIC_BASE_URL is set in the environment — unrelated to this PR, but it bites anyone running the suite from a shell that exports it, and a monkeypatch.delenv would fix it.

No breaking changes: the normalized result gains optional telemetry.degraded / telemetry.error keys and is otherwise unchanged, and normalize_output's new telemetry_error parameter is keyword-only with a default.

Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)

• Relates to NVBug 6562846

• I confirm this contribution is my own work, or I have the right to submit it under this project's license.

• I searched existing issues and open pull requests, and this does not duplicate existing work.

Summary by CodeRabbit

• Bug Fixes

  • Completed agent responses remain successful when telemetry teardown or artifact collection fails.
  • Telemetry issues are reported separately with degraded status and error details.
  • Failures during telemetry setup or before agent execution continue to return invocation errors.
  • Agent execution failures remain invocation errors, while later telemetry failures are reported independently.
  • Partial or unavailable telemetry artifacts no longer replace completed functional responses.

• Documentation

  • Clarified telemetry failure handling across setup, execution, teardown, and artifact collection.

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

Walkthrough

The DeepAgents adapter separates agent execution failures from Relay telemetry setup, teardown, and artifact-collection failures. Completed results remain successful after post-invocation telemetry faults, which are reported through degraded telemetry output.

Changes

Relay telemetry lifecycle

Layer / File(s)	Summary
Separate agent and telemetry failures adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py	Invocation handling tracks agent and telemetry outcomes independently. Setup failures remain invocation errors. Post-invocation teardown and collection failures become telemetry errors.
Expose telemetry errors in output adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py	normalize_output accepts telemetry_error and emits degraded telemetry with combined errors when observability fails.
Validate Relay lifecycle outcomes tests/adapters/test_deepagents.py, adapters/deepagents/README.md	Tests cover setup, execution, teardown, artifact collection, and combined failures. Documentation defines the telemetry failure behavior.

Estimated code review effort: 3 (Moderate) | ~25 minutes

Sequence Diagram(s)

sequenceDiagram
  participant Invocation
  participant Relay
  participant Agent
  participant Artifacts
  participant Output
  Invocation->>Relay: Set up telemetry scope
  Relay->>Agent: Hand off execution
  Agent-->>Invocation: Return result or invocation error
  Invocation->>Relay: Tear down telemetry scope
  Relay-->>Invocation: Return telemetry error when teardown fails
  Invocation->>Artifacts: Collect telemetry artifacts
  Artifacts-->>Invocation: Return artifacts or collection error
  Invocation->>Output: Normalize result and telemetry status

Loading

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 41.38% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title follows Conventional Commits format, uses an allowed lowercase type, states the main fix, and is under 72 characters.
Description check	✅ Passed	The description includes all required template sections, explains the design and testing, and identifies the related issue and confirmations.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@adapters/deepagents/README.md`:
- Around line 211-217: In the telemetry failure description, replace “Once the
agent has been invoked” with “After the agent has been invoked” to follow the
temporal wording guideline while preserving the stated outcome and
error-reporting behavior.

In `@adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py`:
- Around line 612-617: Update the exception handling around telemetry lifecycle
failures in the adapter’s invocation flow so `str(exc)` is never placed in the
returned response; use an allowlisted error code or fixed safe message for
pre-invocation failures while preserving the existing invocation-failure
behavior. Add a regression test using a secret-like exception message and assert
that the lifecycle output does not contain that value.

In `@tests/adapters/test_deepagents.py`:
- Around line 449-468: Update
test_relay_setup_failure_before_the_agent_runs_stays_an_invocation_failure to
replace or stub adapter.invoke_compiled_agent with an AsyncMock before calling
invoke_once, then assert the mock was not awaited afterward. Preserve the
existing output assertions while explicitly verifying relay scope setup failure
prevents agent execution.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 26b8fe7f-d325-4cbd-bbf7-93b1a68a46db

📥 Commits

Reviewing files that changed from the base of the PR and between 9f0881e and b4a54df.

📒 Files selected for processing (3)

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

📜 Review details

⏰ Context from checks skipped due to timeout. (13)

• GitHub Check: Test (Python 3.12, linux-arm64)
• GitHub Check: Test (Python 3.14, linux-arm64)
• GitHub Check: Test (Python 3.12, linux-amd64)
• GitHub Check: Test (Python 3.13, linux-amd64)
• GitHub Check: Test (Python 3.14, windows-amd64)
• GitHub Check: Test (Python 3.13, windows-amd64)
• GitHub Check: Test (Python 3.11, linux-arm64)
• GitHub Check: Test (Python 3.12, windows-amd64)
• GitHub Check: Test (Python 3.11, windows-amd64)
• GitHub Check: Test (Python 3.14, linux-amd64)
• GitHub Check: Test (Python 3.13, linux-arm64)
• GitHub Check: Test (Python 3.11, linux-amd64)
• GitHub Check: Pre-commit

🧰 Additional context used

📓 Path-based instructions (24)

**/*.{md,rst}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Update documentation and examples in the same branch as the public API change.

Files:

• adapters/deepagents/README.md

**/*

📄 CodeRabbit inference engine (.agents/skills/karpathy-guidelines/SKILL.md)

**/*: Before implementing, explicitly state assumptions, surface ambiguity and tradeoffs, present multiple interpretations when relevant, and ask for clarification rather than silently deciding or proceeding when requirements are unclear.
Prefer the minimum code needed to solve the requested problem: avoid speculative features, unnecessary abstractions, unrequested flexibility, and handling of impossible scenarios; simplify overcomplicated solutions.
When editing existing code, make surgical changes only: do not modify unrelated code, comments, formatting, or pre-existing dead code; match the existing style, and remove only unused imports, variables, or functions introduced by your changes.
Define verifiable success criteria for each task, such as writing regression tests for bugs and invalid-input tests for validation, then verify the implementation against those criteria. For multi-step work, state a brief plan with a verification check for each step.

**/*: Always spell NVIDIA in all caps; do not use Nvidia, nvidia, nVidia, nVIDIA, or NV.
Use an NVIDIA before a noun, because the name begins with an “en” sound.
Do not add a registered trademark symbol after NVIDIA when referring to the company; use trademark symbols with product names only when required by the document type or legal guidance.
Verify official capitalization, spacing, hyphenation, and spelling for NVIDIA and third-party product names; do not rewrite official product names for grammar or title-case rules.
Precede NVIDIA product names with NVIDIA on first mention when natural and accurate, and link the first mention when the destination helps the reader.
On first use, include the company name and full model qualifier when it helps identify the model; preserve official capitalization and punctuation, and use shorter family names only after establishing the full name.
For learning-oriented and developer content, do not force trademark symbols unless explicitly required; for press, ...

Files:

• adapters/deepagents/README.md
• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{md,mdx,rst}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-guide.md)

**/*.{md,mdx,rst}: For NeMo Fabric documentation, verify technical claims against the current repository, public API, or documented command before reviewing style.
Always spell NVIDIA in all caps; do not use Nvidia, nvidia, or NV.
Format commands, code elements, expressions, package names, file names, and paths as inline code.
Use descriptive link text; avoid raw URLs and weak anchors such as here or read more.
Use title case consistently for technical documentation headings.
Introduce code blocks, lists, tables, and images with complete sentences.
Write procedures as imperative, parallel steps; split long procedures into smaller tasks.
Prefer active voice, present tense, short sentences, contractions, and plain English while preserving necessary technical precision.
Use can for possibility and reserve may for permission.
Use after for temporal relationships instead of once, and prefer refer to over see when directing readers to another resource.
Avoid culture-specific idioms, unnecessary Latinisms, jokes, and marketing exaggeration in technical documentation.
Spell out months in body text, avoid ordinal dates, and use clear time zones.
Spell out whole numbers from zero through nine unless they are technical values, parameters, versions, or UI values; use numerals for 10 or greater and commas in thousands.
Do not add trademark symbols to learning-oriented documentation unless the source, platform, or legal guidance explicitly requires them.
Do not replace precise technical terms with simpler words when doing so would lose precision.
Do not flag passive voice when the actor is unknown or the action is the important part.
Do not rewrite API names, package names, command flags, or code literals for style.

**/*.{md,mdx,rst}: Use consistent title case for technical-document headings and table headers; avoid quotation marks, ampersands, and exclamation marks in headings, while preserving official product, event, research, and whitepaper title ...

Files:

• adapters/deepagents/README.md

**/*.{md,rst,txt,adoc}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-language-mechanics.md)

**/*.{md,rst,txt,adoc}: For technical documentation, use professional, active, conversational, engaging, precise, and plain-English prose. Prefer active voice, present tense, short sentences, and scannable paragraphs. Avoid casual or imprecise language, swearing, threats, insults, jokes, puns, culture-specific idioms, marketing exaggeration, and unsupported third-party comparisons.
Use can for possibility and reserve may for permission; use after for temporal order; use refer to for cross-references; prefer short direct sentences and specific verbs; avoid unnecessary please in technical documentation.
Prefer active voice when the actor matters. Passive voice is acceptable when the actor is unknown or irrelevant, when the action or result is the focus, or in programmer documentation.
Use natural contractions in conversational technical prose, but do not force them in formal legal copy, API references, or generated text.
Prefer simpler English over Latinisms: use for example or such as instead of e.g., and so on instead of etc., that is instead of i.e., compared to instead of vs., and by, through, or using instead of via. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Use that without commas for essential clauses, and which with commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such as June 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space before a.m. or p.m.; use ET and PT for needed time zones; avoid 24/7; and prefer from 12:30 to 1:00 p.m. for prose ranges.
Format numbers consistently: spell out zero through nine in body text, use numerals for 10 or greater and for technical values, use commas in thousands, do not begin a sentence with a numeral, spell out ordinals, and use numerals consistently within a category wh...

Files:

• adapters/deepagents/README.md

**/*.{md,mdx,yml,py,rs,sh}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

Keep documentation aligned with current NeMo Fabric behavior, repository layout, entry points, commands, package names, APIs, bindings, and support claims.

Files:

• adapters/deepagents/README.md
• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{md,mdx}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

**/*.{md,mdx}: Use the full product name NVIDIA NeMo Fabric on first use, typically in the title and H1; use NeMo Fabric thereafter. Use fabric alone only for the CLI tool and surround it with backticks.
Treat incorrect or stale commands, package names, paths, APIs, support claims, procedures, examples, terminology, or public behavior documentation as blocking issues.
Capitalize NVIDIA correctly and format code, commands, paths, and filenames as inline code where needed.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences; ensure examples match current APIs and build commands.
Use descriptive anchor text, avoid raw URLs and generic labels such as here, and use repository-relative .mdx paths for links within docs/.
Prefer active voice, present tense, short sentences, plain English, consistent terminology, and imperative, parallel, scannable procedures.
Use after instead of once when expressing temporal sequence, and use can rather than may when describing possibility rather than permission.
Avoid ambiguous numeric dates and ordinal dates in body text.
For learning-oriented documentation, do not force trademark symbols unless the source document explicitly requires them.
When reporting documentation-review findings, lead with Must fix, Should fix, and Nice to have categories; include file path, line reference, current problem, rationale, and a concrete rewrite or direction.

Files:

• adapters/deepagents/README.md

**/*.md

📄 CodeRabbit inference engine (CONTRIBUTING.md)

Update relevant SDK, API reference, adapter, example, integration, and support documentation when the corresponding public surface changes.

Files:

• adapters/deepagents/README.md

**/*.{html,md}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

HTML and Markdown files must use the specified SPDX HTML-comment header.

Files:

• adapters/deepagents/README.md

adapters/*/{README.md,fabric-adapter.json}

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Document installation, supported configuration, harness-only settings, credentials, lifecycle, telemetry, artifacts, limitations, and focused test commands; keep documentation consistent with descriptor claims.

Files:

• adapters/deepagents/README.md

adapters/*/README.md

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Provide a credential-free fixture covering plan, doctor, and run, plus a canonical typed SDK example and, when required, a canonical harness-native YAML fixture.

Files:

• adapters/deepagents/README.md

{adapters/**,examples/**}

⚙️ CodeRabbit configuration file

{adapters/**,examples/**}: Review adapter and example changes for command correctness, config/schema consistency, artifact handling, and compatibility with the public NeMo Fabric contracts.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

{*.md,**/*.md,**/*.mdx,**/*.ipynb}

⚙️ CodeRabbit configuration file

{*.md,**/*.md,**/*.mdx,**/*.ipynb}: Enforce the product name in user-facing prose: use "NVIDIA NeMo Fabric" on first use and "NeMo Fabric" thereafter. Flag standalone capitalized "Fabric" when it refers to the product. Do not flag the lowercase fabric CLI command, package/import/crate names, code identifiers, API symbols, configuration keys, file paths, or unrelated generic uses of the word.

Files:

• adapters/deepagents/README.md

**/*.{rs,py,pyi,json,yaml,yml}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Determine and update every affected public surface, including the CLI, PyO3 bindings, Python SDK, type stubs, schemas, and adapter contract, so they remain in parity.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{rs,py}

📄 CodeRabbit inference engine (.agents/skills/prepare-pr/SKILL.md)

For native binding changes, run cargo check -p fabric-python --locked.

Use snake_case for functions and variables; use PascalCase for Rust types and Python classes.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

If Python code or a Python-facing adapter changes, run just test-python.

In Python SDK, adapters, examples, and tests, follow the existing style, use type annotations for public APIs, and keep native binding declarations synchronized with their Rust implementations.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{rs,py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

**/*.{rs,py,pyi}: If public configuration types change, confirm schema snapshot tests in just test-rust pass and review generated schema diffs.
For schema or public contract changes, run both language suites and review changes under schemas/ and generated API references.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

tests/adapters/**/*.py

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

tests/adapters/**/*.py: If an adapter or integration changes, run its focused tests.
For adapter behavior changes, run focused adapter tests under tests/adapters, then run just test-python.

Files:

• tests/adapters/test_deepagents.py

**/*.{py,pyi,rs}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

For Python SDK or PyO3 binding changes, use python-tests, run focused pytest tests first, then just test-python; rebuild with just build-python when native code or packaging changes.

Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{rs,py,toml}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

When editing version helpers, verify every nemo-fabric-* workspace package through Cargo metadata and reject a static version in python/pyproject.toml.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

**/*.{toml,rs,py}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

Avoid blind repository-wide replacement of version-like strings; distinguish package-version references from examples and unrelated dependency versions.

Files:

• tests/adapters/test_deepagents.py
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

tests/**/*.{rs,py}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

When adding functionality, include tests in the corresponding Rust crate or the relevant area under tests/.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.py

📄 CodeRabbit inference engine (.agents/skills/python-tests/SKILL.md)

tests/**/*.py: Use pytest to run Python tests.
Do not add @pytest.mark.asyncio to tests; async tests are automatically detected by the async runner.
Do not add -> None return type annotations to test functions.
When mocking a class, use unittest.mock.MagicMock or AsyncMock, using the spec argument when necessary, rather than defining a new class.
Prefix mocked class names with mock, not fake.
Prefer pytest fixtures over helper methods.
If a fixture is needed in multiple test files, define it once in conftest.py rather than repeating it.
Define fixtures using @pytest.fixture(name="<fixture_name>"[, scope="<scope>"]) and a <fixture_name>_fixture function; specify scope only when it is not function.
Prefer pytest.mark.parametrize over separate tests for different input types.
Use @pytest.mark.usefixtures when a fixture is needed but its returned value is unused or it returns no value.
Avoid defensive programming in tests; access expected values directly so missing data raises a clear failure, such as using results["data"] instead of results.get("data").
When adapter installation metadata changes, packaging metadata tests must directly assert that the root project depends unconditionally on the exact-version nemo-fabric-runtime distribution.
Packaging metadata tests must verify that each root harness extra delegates to the matching version of the leaf adapter's harness extra.
Packaging metadata tests must verify that bare leaf dependencies remain adapter-owned and that the root adapter-tests dependency group installs each leaf through its harness extra.
Packaging metadata tests must verify that every leaf provides full; only adapters importing NeMo Relay Python APIs provide relay, while adapters using an external Relay executable have full equal to harness.

Files:

• tests/adapters/test_deepagents.py

{tests/**,python/tests/**}

⚙️ CodeRabbit configuration file

{tests/**,python/tests/**}: Tests should cover the behavior promised by the changed API surface, including error paths, lifecycle cleanup, and SDK/native parity where relevant.

Files:

• tests/adapters/test_deepagents.py

adapters/*/src/**/*.py

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

adapters/*/src/**/*.py: Implement adapters using the existing Fabric python or process runner and normalized request/result contracts; do not add a runner or one-off abstraction for a single adapter.
Treat normalized config, Fabric-resolved plans, and runtime_context as authoritative; reserve harness.settings for adapter-wide behavior and apply precedence in the order: normalized config, plans/context, harness settings, descriptor/default values.
Reject conflicting duplicate declarations and unsupported behavior with actionable errors naming the field and supported alternatives; never silently drop configuration.
Validate dependency versions, hooks, and credentials before harness invocation, and never expose credential values in outputs, errors, events, logs, or fixtures.
Forward only required system, selected credential, telemetry, and documented harness-specific environment variables; never forward or log unrelated environment values.
Maintain one local adapter host per Fabric runtime across ordered start → invoke* → stop operations. Emit one JSON lifecycle response per request on stdout and diagnostics on stderr.
Return harness-level invocation failures as successful lifecycle responses containing response: null, failed: true, and structured error fields (code, message, retryable, and optional metadata).
Do not emit NeMo Relay stream records on adapter stdout; return exactly one terminal lifecycle response while streaming occurs through the SDK-owned out-of-band endpoint.
Scope workspace, generated configuration, state, sessions, and artifacts to the resolved runtime context, and isolate stateful adapter instances by Fabric runtime ID.
Use start to initialize adapter-owned harness state, retain it for continuation across repeated invoke calls on the same runtime, and release it in stop.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

🧠 Learnings (1)

📚 Learning: 2026-07-09T22:28:51.689Z

Learnt from: AjayThorve
Repo: NVIDIA/NeMo-Fabric PR: 43
File: adapters/claude-sdk/src/nemo_fabric_adapters/claude_sdk/adapter.py:164-168
Timestamp: 2026-07-09T22:28:51.689Z
Learning: In the NeMo-Fabric adapters, treat path values used in Fabric adapter configuration (including logic like `_resolve_path` in adapter.py) as config-root-relative. Do not apply `Path.expanduser()` (or otherwise apply `~`/home or shell-style expansion), because it will make the resolved paths normalize inconsistently across adapters. Also, do not rely on or add any resolution behavior that uses `harness.settings.cwd` as an override point for these adapter paths—`harness.settings.cwd` is explicitly unsupported in this adapter context.

Applied to files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

🪛 Ruff (0.16.1)

tests/adapters/test_deepagents.py

[warning] 416-416: Unused function argument: fake_sdks

(ARG001)

---

[warning] 416-416: Unused function argument: fake_relay

(ARG001)

---

[warning] 428-428: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 428-428: Unused function argument: name

(ARG001)

---

[warning] 428-428: Unused function argument: scope_type

(ARG001)

---

[warning] 428-428: Unused function argument: kwargs

(ARG001)

---

[warning] 430-432: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 450-450: Unused function argument: fake_sdks

(ARG001)

---

[warning] 450-450: Unused function argument: fake_relay

(ARG001)

---

[warning] 457-457: Missing return type annotation for private function failing_scope

(ANN202)

---

[warning] 457-457: Unused function argument: name

(ARG001)

---

[warning] 457-457: Unused function argument: scope_type

(ARG001)

---

[warning] 457-457: Unused function argument: kwargs

(ARG001)

---

[warning] 458-458: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 471-471: Unused function argument: fake_sdks

(ARG001)

---

[warning] 471-471: Unused function argument: fake_relay

(ARG001)

---

[warning] 475-475: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 476-476: Avoid specifying long messages outside the exception class

(TRY003)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

[warning] 612-612: Do not catch blind exception: Exception

(BLE001)

---

[warning] 656-656: Do not catch blind exception: Exception

(BLE001)

🔇 Additional comments (3)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py (1)

638-658: LGTM!

tests/adapters/test_deepagents.py (2)

389-412: LGTM!

Also applies to: 415-447

---

470-487: LGTM!

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@adapters/deepagents/README.md`:
- Around line 211-217: Update the telemetry behavior description to qualify the
completed-result claim: telemetry faults preserve a completed result only when
agent execution succeeds, while an agent invocation failure remains failed even
if telemetry teardown or reporting also fails. Retain the existing description
of recording and joining telemetry faults.

In `@adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py`:
- Around line 616-629: Move callback handler construction from before the try
block into the try block surrounding the telemetry and agent invocation flow in
invoke. Ensure constructor exceptions are caught by the existing telemetry
lifecycle error handling and converted into the required invocation failure
response, and add a test covering a callback handler constructor failure.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 250ad950-2a27-47bb-949f-d0ff175aa11a

📥 Commits

Reviewing files that changed from the base of the PR and between b4a54df and 1a0da88.

📒 Files selected for processing (3)

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

📜 Review details

⏰ Context from checks skipped due to timeout. (17)

• GitHub Check: Pre-commit
• GitHub Check: Test (Python 3.13, windows-amd64)
• GitHub Check: Test (Python 3.12, windows-amd64)
• GitHub Check: Test (Python 3.14, windows-amd64)
• GitHub Check: Test (Python 3.11, linux-amd64)
• GitHub Check: Test (Python 3.14, linux-amd64)
• GitHub Check: Test (Python 3.11, windows-amd64)
• GitHub Check: Test (Python 3.13, linux-arm64)
• GitHub Check: Test (Python 3.12, macos-arm64)
• GitHub Check: Test (Python 3.14, linux-arm64)
• GitHub Check: Test (Python 3.12, linux-amd64)
• GitHub Check: Test (Python 3.13, linux-amd64)
• GitHub Check: Test (Python 3.11, macos-arm64)
• GitHub Check: Test (Python 3.14, macos-arm64)
• GitHub Check: Test (Python 3.11, linux-arm64)
• GitHub Check: Test (Python 3.13, macos-arm64)
• GitHub Check: Test (Python 3.12, linux-arm64)

🧰 Additional context used

📓 Path-based instructions (24)

**/*.{md,rst}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Update documentation and examples in the same branch as the public API change.

Files:

• adapters/deepagents/README.md

**/*

📄 CodeRabbit inference engine (.agents/skills/karpathy-guidelines/SKILL.md)

**/*: Before implementing, explicitly state assumptions, surface ambiguity and tradeoffs, present multiple interpretations when relevant, and ask for clarification rather than silently deciding or proceeding when requirements are unclear.
Prefer the minimum code needed to solve the requested problem: avoid speculative features, unnecessary abstractions, unrequested flexibility, and handling of impossible scenarios; simplify overcomplicated solutions.
When editing existing code, make surgical changes only: do not modify unrelated code, comments, formatting, or pre-existing dead code; match the existing style, and remove only unused imports, variables, or functions introduced by your changes.
Define verifiable success criteria for each task, such as writing regression tests for bugs and invalid-input tests for validation, then verify the implementation against those criteria. For multi-step work, state a brief plan with a verification check for each step.

**/*: Always spell NVIDIA in all caps; do not use Nvidia, nvidia, nVidia, nVIDIA, or NV.
Use an NVIDIA before a noun, because the name begins with an “en” sound.
Do not add a registered trademark symbol after NVIDIA when referring to the company; use trademark symbols with product names only when required by the document type or legal guidance.
Verify official capitalization, spacing, hyphenation, and spelling for NVIDIA and third-party product names; do not rewrite official product names for grammar or title-case rules.
Precede NVIDIA product names with NVIDIA on first mention when natural and accurate, and link the first mention when the destination helps the reader.
On first use, include the company name and full model qualifier when it helps identify the model; preserve official capitalization and punctuation, and use shorter family names only after establishing the full name.
For learning-oriented and developer content, do not force trademark symbols unless explicitly required; for press, ...

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{md,mdx,rst}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-guide.md)

**/*.{md,mdx,rst}: For NeMo Fabric documentation, verify technical claims against the current repository, public API, or documented command before reviewing style.
Always spell NVIDIA in all caps; do not use Nvidia, nvidia, or NV.
Format commands, code elements, expressions, package names, file names, and paths as inline code.
Use descriptive link text; avoid raw URLs and weak anchors such as here or read more.
Use title case consistently for technical documentation headings.
Introduce code blocks, lists, tables, and images with complete sentences.
Write procedures as imperative, parallel steps; split long procedures into smaller tasks.
Prefer active voice, present tense, short sentences, contractions, and plain English while preserving necessary technical precision.
Use can for possibility and reserve may for permission.
Use after for temporal relationships instead of once, and prefer refer to over see when directing readers to another resource.
Avoid culture-specific idioms, unnecessary Latinisms, jokes, and marketing exaggeration in technical documentation.
Spell out months in body text, avoid ordinal dates, and use clear time zones.
Spell out whole numbers from zero through nine unless they are technical values, parameters, versions, or UI values; use numerals for 10 or greater and commas in thousands.
Do not add trademark symbols to learning-oriented documentation unless the source, platform, or legal guidance explicitly requires them.
Do not replace precise technical terms with simpler words when doing so would lose precision.
Do not flag passive voice when the actor is unknown or the action is the important part.
Do not rewrite API names, package names, command flags, or code literals for style.

**/*.{md,mdx,rst}: Use consistent title case for technical-document headings and table headers; avoid quotation marks, ampersands, and exclamation marks in headings, while preserving official product, event, research, and whitepaper title ...

Files:

• adapters/deepagents/README.md

**/*.{md,rst,txt,adoc}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-language-mechanics.md)

**/*.{md,rst,txt,adoc}: For technical documentation, use professional, active, conversational, engaging, precise, and plain-English prose. Prefer active voice, present tense, short sentences, and scannable paragraphs. Avoid casual or imprecise language, swearing, threats, insults, jokes, puns, culture-specific idioms, marketing exaggeration, and unsupported third-party comparisons.
Use can for possibility and reserve may for permission; use after for temporal order; use refer to for cross-references; prefer short direct sentences and specific verbs; avoid unnecessary please in technical documentation.
Prefer active voice when the actor matters. Passive voice is acceptable when the actor is unknown or irrelevant, when the action or result is the focus, or in programmer documentation.
Use natural contractions in conversational technical prose, but do not force them in formal legal copy, API references, or generated text.
Prefer simpler English over Latinisms: use for example or such as instead of e.g., and so on instead of etc., that is instead of i.e., compared to instead of vs., and by, through, or using instead of via. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Use that without commas for essential clauses, and which with commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such as June 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space before a.m. or p.m.; use ET and PT for needed time zones; avoid 24/7; and prefer from 12:30 to 1:00 p.m. for prose ranges.
Format numbers consistently: spell out zero through nine in body text, use numerals for 10 or greater and for technical values, use commas in thousands, do not begin a sentence with a numeral, spell out ordinals, and use numerals consistently within a category wh...

Files:

• adapters/deepagents/README.md

**/*.{md,mdx,yml,py,rs,sh}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

Keep documentation aligned with current NeMo Fabric behavior, repository layout, entry points, commands, package names, APIs, bindings, and support claims.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{md,mdx}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

**/*.{md,mdx}: Use the full product name NVIDIA NeMo Fabric on first use, typically in the title and H1; use NeMo Fabric thereafter. Use fabric alone only for the CLI tool and surround it with backticks.
Treat incorrect or stale commands, package names, paths, APIs, support claims, procedures, examples, terminology, or public behavior documentation as blocking issues.
Capitalize NVIDIA correctly and format code, commands, paths, and filenames as inline code where needed.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences; ensure examples match current APIs and build commands.
Use descriptive anchor text, avoid raw URLs and generic labels such as here, and use repository-relative .mdx paths for links within docs/.
Prefer active voice, present tense, short sentences, plain English, consistent terminology, and imperative, parallel, scannable procedures.
Use after instead of once when expressing temporal sequence, and use can rather than may when describing possibility rather than permission.
Avoid ambiguous numeric dates and ordinal dates in body text.
For learning-oriented documentation, do not force trademark symbols unless the source document explicitly requires them.
When reporting documentation-review findings, lead with Must fix, Should fix, and Nice to have categories; include file path, line reference, current problem, rationale, and a concrete rewrite or direction.

Files:

• adapters/deepagents/README.md

**/*.md

📄 CodeRabbit inference engine (CONTRIBUTING.md)

Update relevant SDK, API reference, adapter, example, integration, and support documentation when the corresponding public surface changes.

Files:

• adapters/deepagents/README.md

**/*.{html,md}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

HTML and Markdown files must use the specified SPDX HTML-comment header.

Files:

• adapters/deepagents/README.md

adapters/*/{README.md,fabric-adapter.json}

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Document installation, supported configuration, harness-only settings, credentials, lifecycle, telemetry, artifacts, limitations, and focused test commands; keep documentation consistent with descriptor claims.

Files:

• adapters/deepagents/README.md

adapters/*/README.md

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Provide a credential-free fixture covering plan, doctor, and run, plus a canonical typed SDK example and, when required, a canonical harness-native YAML fixture.

Files:

• adapters/deepagents/README.md

{adapters/**,examples/**}

⚙️ CodeRabbit configuration file

{adapters/**,examples/**}: Review adapter and example changes for command correctness, config/schema consistency, artifact handling, and compatibility with the public NeMo Fabric contracts.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

{*.md,**/*.md,**/*.mdx,**/*.ipynb}

⚙️ CodeRabbit configuration file

{*.md,**/*.md,**/*.mdx,**/*.ipynb}: Enforce the product name in user-facing prose: use "NVIDIA NeMo Fabric" on first use and "NeMo Fabric" thereafter. Flag standalone capitalized "Fabric" when it refers to the product. Do not flag the lowercase fabric CLI command, package/import/crate names, code identifiers, API symbols, configuration keys, file paths, or unrelated generic uses of the word.

Files:

• adapters/deepagents/README.md

**/*.{rs,py,pyi,json,yaml,yml}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Determine and update every affected public surface, including the CLI, PyO3 bindings, Python SDK, type stubs, schemas, and adapter contract, so they remain in parity.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py}

📄 CodeRabbit inference engine (.agents/skills/prepare-pr/SKILL.md)

For native binding changes, run cargo check -p fabric-python --locked.

Use snake_case for functions and variables; use PascalCase for Rust types and Python classes.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

If Python code or a Python-facing adapter changes, run just test-python.

In Python SDK, adapters, examples, and tests, follow the existing style, use type annotations for public APIs, and keep native binding declarations synchronized with their Rust implementations.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

**/*.{rs,py,pyi}: If public configuration types change, confirm schema snapshot tests in just test-rust pass and review generated schema diffs.
For schema or public contract changes, run both language suites and review changes under schemas/ and generated API references.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{py,pyi,rs}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

For Python SDK or PyO3 binding changes, use python-tests, run focused pytest tests first, then just test-python; rebuild with just build-python when native code or packaging changes.

Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py,toml}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

When editing version helpers, verify every nemo-fabric-* workspace package through Cargo metadata and reject a static version in python/pyproject.toml.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{toml,rs,py}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

Avoid blind repository-wide replacement of version-like strings; distinguish package-version references from examples and unrelated dependency versions.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

adapters/*/src/**/*.py

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

adapters/*/src/**/*.py: Implement adapters using the existing Fabric python or process runner and normalized request/result contracts; do not add a runner or one-off abstraction for a single adapter.
Treat normalized config, Fabric-resolved plans, and runtime_context as authoritative; reserve harness.settings for adapter-wide behavior and apply precedence in the order: normalized config, plans/context, harness settings, descriptor/default values.
Reject conflicting duplicate declarations and unsupported behavior with actionable errors naming the field and supported alternatives; never silently drop configuration.
Validate dependency versions, hooks, and credentials before harness invocation, and never expose credential values in outputs, errors, events, logs, or fixtures.
Forward only required system, selected credential, telemetry, and documented harness-specific environment variables; never forward or log unrelated environment values.
Maintain one local adapter host per Fabric runtime across ordered start → invoke* → stop operations. Emit one JSON lifecycle response per request on stdout and diagnostics on stderr.
Return harness-level invocation failures as successful lifecycle responses containing response: null, failed: true, and structured error fields (code, message, retryable, and optional metadata).
Do not emit NeMo Relay stream records on adapter stdout; return exactly one terminal lifecycle response while streaming occurs through the SDK-owned out-of-band endpoint.
Scope workspace, generated configuration, state, sessions, and artifacts to the resolved runtime context, and isolate stateful adapter instances by Fabric runtime ID.
Use start to initialize adapter-owned harness state, retain it for continuation across repeated invoke calls on the same runtime, and release it in stop.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

tests/adapters/**/*.py

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

tests/adapters/**/*.py: If an adapter or integration changes, run its focused tests.
For adapter behavior changes, run focused adapter tests under tests/adapters, then run just test-python.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.{rs,py}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

When adding functionality, include tests in the corresponding Rust crate or the relevant area under tests/.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.py

📄 CodeRabbit inference engine (.agents/skills/python-tests/SKILL.md)

tests/**/*.py: Use pytest to run Python tests.
Do not add @pytest.mark.asyncio to tests; async tests are automatically detected by the async runner.
Do not add -> None return type annotations to test functions.
When mocking a class, use unittest.mock.MagicMock or AsyncMock, using the spec argument when necessary, rather than defining a new class.
Prefix mocked class names with mock, not fake.
Prefer pytest fixtures over helper methods.
If a fixture is needed in multiple test files, define it once in conftest.py rather than repeating it.
Define fixtures using @pytest.fixture(name="<fixture_name>"[, scope="<scope>"]) and a <fixture_name>_fixture function; specify scope only when it is not function.
Prefer pytest.mark.parametrize over separate tests for different input types.
Use @pytest.mark.usefixtures when a fixture is needed but its returned value is unused or it returns no value.
Avoid defensive programming in tests; access expected values directly so missing data raises a clear failure, such as using results["data"] instead of results.get("data").
When adapter installation metadata changes, packaging metadata tests must directly assert that the root project depends unconditionally on the exact-version nemo-fabric-runtime distribution.
Packaging metadata tests must verify that each root harness extra delegates to the matching version of the leaf adapter's harness extra.
Packaging metadata tests must verify that bare leaf dependencies remain adapter-owned and that the root adapter-tests dependency group installs each leaf through its harness extra.
Packaging metadata tests must verify that every leaf provides full; only adapters importing NeMo Relay Python APIs provide relay, while adapters using an external Relay executable have full equal to harness.

Files:

• tests/adapters/test_deepagents.py

{tests/**,python/tests/**}

⚙️ CodeRabbit configuration file

{tests/**,python/tests/**}: Tests should cover the behavior promised by the changed API surface, including error paths, lifecycle cleanup, and SDK/native parity where relevant.

Files:

• tests/adapters/test_deepagents.py

🧠 Learnings (1)

📚 Learning: 2026-07-09T22:28:51.689Z

Learnt from: AjayThorve
Repo: NVIDIA/NeMo-Fabric PR: 43
File: adapters/claude-sdk/src/nemo_fabric_adapters/claude_sdk/adapter.py:164-168
Timestamp: 2026-07-09T22:28:51.689Z
Learning: In the NeMo-Fabric adapters, treat path values used in Fabric adapter configuration (including logic like `_resolve_path` in adapter.py) as config-root-relative. Do not apply `Path.expanduser()` (or otherwise apply `~`/home or shell-style expansion), because it will make the resolved paths normalize inconsistently across adapters. Also, do not rely on or add any resolution behavior that uses `harness.settings.cwd` as an override point for these adapter paths—`harness.settings.cwd` is explicitly unsupported in this adapter context.

Applied to files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

🪛 Ruff (0.16.1)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

[warning] 629-629: Do not catch blind exception: Exception

(BLE001)

---

[warning] 659-659: Do not catch blind exception: Exception

(BLE001)

---

[warning] 691-691: Do not catch blind exception: Exception

(BLE001)

tests/adapters/test_deepagents.py

[warning] 416-416: Unused function argument: fake_sdks

(ARG001)

---

[warning] 416-416: Unused function argument: fake_relay

(ARG001)

---

[warning] 428-428: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 428-428: Unused function argument: name

(ARG001)

---

[warning] 428-428: Unused function argument: scope_type

(ARG001)

---

[warning] 428-428: Unused function argument: kwargs

(ARG001)

---

[warning] 430-432: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 451-451: Unused function argument: fake_sdks

(ARG001)

---

[warning] 451-451: Unused function argument: fake_relay

(ARG001)

---

[warning] 462-462: Missing return type annotation for private function exploding_plugin

(ANN202)

---

[warning] 462-462: Unused function argument: config

(ARG001)

---

[warning] 464-464: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 479-479: Unused function argument: fake_sdks

(ARG001)

---

[warning] 479-479: Unused function argument: fake_relay

(ARG001)

---

[warning] 488-488: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 502-502: Unused function argument: fake_sdks

(ARG001)

---

[warning] 502-502: Unused function argument: fake_relay

(ARG001)

---

[warning] 509-509: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 509-509: Unused function argument: name

(ARG001)

---

[warning] 509-509: Unused function argument: scope_type

(ARG001)

---

[warning] 509-509: Unused function argument: kwargs

(ARG001)

---

[warning] 511-511: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 514-514: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 528-528: Unused function argument: fake_sdks

(ARG001)

---

[warning] 528-528: Unused function argument: fake_relay

(ARG001)

---

[warning] 535-535: Missing return type annotation for private function failing_scope

(ANN202)

---

[warning] 535-535: Unused function argument: name

(ARG001)

---

[warning] 535-535: Unused function argument: scope_type

(ARG001)

---

[warning] 535-535: Unused function argument: kwargs

(ARG001)

---

[warning] 536-536: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 552-552: Unused function argument: fake_sdks

(ARG001)

---

[warning] 552-552: Unused function argument: fake_relay

(ARG001)

---

[warning] 556-556: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 557-557: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 572-572: Unused function argument: fake_sdks

(ARG001)

---

[warning] 572-572: Unused function argument: fake_relay

(ARG001)

---

[warning] 578-578: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 579-579: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 582-582: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 582-582: Unused function argument: name

(ARG001)

---

[warning] 582-582: Unused function argument: scope_type

(ARG001)

---

[warning] 582-582: Unused function argument: kwargs

(ARG001)

---

[warning] 584-584: Avoid specifying long messages outside the exception class

(TRY003)

🔇 Additional comments (4)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py (2)

809-810: Do Not Return Raw Exception Text.

_error_text() places arbitrary exception text into lifecycle output. Exceptions can include plugin configuration or credential values. This remains unresolved from the earlier review finding.

As per coding guidelines, “never expose credential values in outputs, errors, events, logs, or fixtures.”

Source: Coding guidelines

---

573-595: LGTM!

tests/adapters/test_deepagents.py (2)

527-548: Assert That Scope Setup Blocks Agent Execution.

The test checks the returned error, but it does not prove that the agent was not called. Stub adapter.invoke_compiled_agent with AsyncMock and assert that it was not awaited.

As per path instructions, “Tests should cover the behavior promised by the changed API surface.”

Source: Path instructions

---

389-412: LGTM!

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@adapters/deepagents/README.md`:
- Around line 220-224: Update the documentation around the telemetry.degraded
artifact behavior to distinguish teardown failures from artifact-collection
failures: artifacts may still be referenced when teardown does not complete
cleanly, but are unavailable when collection fails. Replace the unconditional
degraded-run artifact claim while preserving the guidance to treat degraded
trajectories as untrusted.

In `@adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py`:
- Around line 612-629: Update the lifecycle handling around _relay_plugin.plugin
and _relay_scope.scope so scope-exit and plugin __aexit__ exceptions are
captured independently rather than allowing the later plugin failure to replace
the scope failure. Combine both telemetry faults with _join_faults, preserve the
existing outcome behavior, and add a test covering simultaneous failures from
both teardown operations.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 33c8c148-5080-4f57-a6d1-0df15494c9f4

📥 Commits

Reviewing files that changed from the base of the PR and between 1a0da88 and b33a174.

📒 Files selected for processing (3)

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

📜 Review details

⏰ Context from checks skipped due to timeout. (17)

• GitHub Check: Test (Python 3.14, linux-amd64)
• GitHub Check: Test (Python 3.14, windows-amd64)
• GitHub Check: Test (Python 3.13, linux-amd64)
• GitHub Check: Test (Python 3.14, macos-arm64)
• GitHub Check: Test (Python 3.11, windows-amd64)
• GitHub Check: Test (Python 3.13, macos-arm64)
• GitHub Check: Test (Python 3.13, linux-arm64)
• GitHub Check: Test (Python 3.12, macos-arm64)
• GitHub Check: Test (Python 3.11, macos-arm64)
• GitHub Check: Test (Python 3.13, windows-amd64)
• GitHub Check: Test (Python 3.14, linux-arm64)
• GitHub Check: Test (Python 3.12, linux-arm64)
• GitHub Check: Test (Python 3.12, windows-amd64)
• GitHub Check: Test (Python 3.12, linux-amd64)
• GitHub Check: Test (Python 3.11, linux-amd64)
• GitHub Check: Test (Python 3.11, linux-arm64)
• GitHub Check: Pre-commit

🧰 Additional context used

📓 Path-based instructions (24)

**/*.{md,rst}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Update documentation and examples in the same branch as the public API change.

Files:

• adapters/deepagents/README.md

**/*

📄 CodeRabbit inference engine (.agents/skills/karpathy-guidelines/SKILL.md)

**/*: Before implementing, explicitly state assumptions, surface ambiguity and tradeoffs, present multiple interpretations when relevant, and ask for clarification rather than silently deciding or proceeding when requirements are unclear.
Prefer the minimum code needed to solve the requested problem: avoid speculative features, unnecessary abstractions, unrequested flexibility, and handling of impossible scenarios; simplify overcomplicated solutions.
When editing existing code, make surgical changes only: do not modify unrelated code, comments, formatting, or pre-existing dead code; match the existing style, and remove only unused imports, variables, or functions introduced by your changes.
Define verifiable success criteria for each task, such as writing regression tests for bugs and invalid-input tests for validation, then verify the implementation against those criteria. For multi-step work, state a brief plan with a verification check for each step.

**/*: Always spell NVIDIA in all caps; do not use Nvidia, nvidia, nVidia, nVIDIA, or NV.
Use an NVIDIA before a noun, because the name begins with an “en” sound.
Do not add a registered trademark symbol after NVIDIA when referring to the company; use trademark symbols with product names only when required by the document type or legal guidance.
Verify official capitalization, spacing, hyphenation, and spelling for NVIDIA and third-party product names; do not rewrite official product names for grammar or title-case rules.
Precede NVIDIA product names with NVIDIA on first mention when natural and accurate, and link the first mention when the destination helps the reader.
On first use, include the company name and full model qualifier when it helps identify the model; preserve official capitalization and punctuation, and use shorter family names only after establishing the full name.
For learning-oriented and developer content, do not force trademark symbols unless explicitly required; for press, ...

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{md,mdx,rst}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-guide.md)

**/*.{md,mdx,rst}: For NeMo Fabric documentation, verify technical claims against the current repository, public API, or documented command before reviewing style.
Always spell NVIDIA in all caps; do not use Nvidia, nvidia, or NV.
Format commands, code elements, expressions, package names, file names, and paths as inline code.
Use descriptive link text; avoid raw URLs and weak anchors such as here or read more.
Use title case consistently for technical documentation headings.
Introduce code blocks, lists, tables, and images with complete sentences.
Write procedures as imperative, parallel steps; split long procedures into smaller tasks.
Prefer active voice, present tense, short sentences, contractions, and plain English while preserving necessary technical precision.
Use can for possibility and reserve may for permission.
Use after for temporal relationships instead of once, and prefer refer to over see when directing readers to another resource.
Avoid culture-specific idioms, unnecessary Latinisms, jokes, and marketing exaggeration in technical documentation.
Spell out months in body text, avoid ordinal dates, and use clear time zones.
Spell out whole numbers from zero through nine unless they are technical values, parameters, versions, or UI values; use numerals for 10 or greater and commas in thousands.
Do not add trademark symbols to learning-oriented documentation unless the source, platform, or legal guidance explicitly requires them.
Do not replace precise technical terms with simpler words when doing so would lose precision.
Do not flag passive voice when the actor is unknown or the action is the important part.
Do not rewrite API names, package names, command flags, or code literals for style.

**/*.{md,mdx,rst}: Use consistent title case for technical-document headings and table headers; avoid quotation marks, ampersands, and exclamation marks in headings, while preserving official product, event, research, and whitepaper title ...

Files:

• adapters/deepagents/README.md

**/*.{md,rst,txt,adoc}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-language-mechanics.md)

**/*.{md,rst,txt,adoc}: For technical documentation, use professional, active, conversational, engaging, precise, and plain-English prose. Prefer active voice, present tense, short sentences, and scannable paragraphs. Avoid casual or imprecise language, swearing, threats, insults, jokes, puns, culture-specific idioms, marketing exaggeration, and unsupported third-party comparisons.
Use can for possibility and reserve may for permission; use after for temporal order; use refer to for cross-references; prefer short direct sentences and specific verbs; avoid unnecessary please in technical documentation.
Prefer active voice when the actor matters. Passive voice is acceptable when the actor is unknown or irrelevant, when the action or result is the focus, or in programmer documentation.
Use natural contractions in conversational technical prose, but do not force them in formal legal copy, API references, or generated text.
Prefer simpler English over Latinisms: use for example or such as instead of e.g., and so on instead of etc., that is instead of i.e., compared to instead of vs., and by, through, or using instead of via. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Use that without commas for essential clauses, and which with commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such as June 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space before a.m. or p.m.; use ET and PT for needed time zones; avoid 24/7; and prefer from 12:30 to 1:00 p.m. for prose ranges.
Format numbers consistently: spell out zero through nine in body text, use numerals for 10 or greater and for technical values, use commas in thousands, do not begin a sentence with a numeral, spell out ordinals, and use numerals consistently within a category wh...

Files:

• adapters/deepagents/README.md

**/*.{md,mdx,yml,py,rs,sh}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

Keep documentation aligned with current NeMo Fabric behavior, repository layout, entry points, commands, package names, APIs, bindings, and support claims.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{md,mdx}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

**/*.{md,mdx}: Use the full product name NVIDIA NeMo Fabric on first use, typically in the title and H1; use NeMo Fabric thereafter. Use fabric alone only for the CLI tool and surround it with backticks.
Treat incorrect or stale commands, package names, paths, APIs, support claims, procedures, examples, terminology, or public behavior documentation as blocking issues.
Capitalize NVIDIA correctly and format code, commands, paths, and filenames as inline code where needed.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences; ensure examples match current APIs and build commands.
Use descriptive anchor text, avoid raw URLs and generic labels such as here, and use repository-relative .mdx paths for links within docs/.
Prefer active voice, present tense, short sentences, plain English, consistent terminology, and imperative, parallel, scannable procedures.
Use after instead of once when expressing temporal sequence, and use can rather than may when describing possibility rather than permission.
Avoid ambiguous numeric dates and ordinal dates in body text.
For learning-oriented documentation, do not force trademark symbols unless the source document explicitly requires them.
When reporting documentation-review findings, lead with Must fix, Should fix, and Nice to have categories; include file path, line reference, current problem, rationale, and a concrete rewrite or direction.

Files:

• adapters/deepagents/README.md

**/*.md

📄 CodeRabbit inference engine (CONTRIBUTING.md)

Update relevant SDK, API reference, adapter, example, integration, and support documentation when the corresponding public surface changes.

Files:

• adapters/deepagents/README.md

**/*.{html,md}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

HTML and Markdown files must use the specified SPDX HTML-comment header.

Files:

• adapters/deepagents/README.md

adapters/*/{README.md,fabric-adapter.json}

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Document installation, supported configuration, harness-only settings, credentials, lifecycle, telemetry, artifacts, limitations, and focused test commands; keep documentation consistent with descriptor claims.

Files:

• adapters/deepagents/README.md

adapters/*/README.md

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Provide a credential-free fixture covering plan, doctor, and run, plus a canonical typed SDK example and, when required, a canonical harness-native YAML fixture.

Files:

• adapters/deepagents/README.md

{adapters/**,examples/**}

⚙️ CodeRabbit configuration file

{adapters/**,examples/**}: Review adapter and example changes for command correctness, config/schema consistency, artifact handling, and compatibility with the public NeMo Fabric contracts.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

{*.md,**/*.md,**/*.mdx,**/*.ipynb}

⚙️ CodeRabbit configuration file

{*.md,**/*.md,**/*.mdx,**/*.ipynb}: Enforce the product name in user-facing prose: use "NVIDIA NeMo Fabric" on first use and "NeMo Fabric" thereafter. Flag standalone capitalized "Fabric" when it refers to the product. Do not flag the lowercase fabric CLI command, package/import/crate names, code identifiers, API symbols, configuration keys, file paths, or unrelated generic uses of the word.

Files:

• adapters/deepagents/README.md

**/*.{rs,py,pyi,json,yaml,yml}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Determine and update every affected public surface, including the CLI, PyO3 bindings, Python SDK, type stubs, schemas, and adapter contract, so they remain in parity.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py}

📄 CodeRabbit inference engine (.agents/skills/prepare-pr/SKILL.md)

For native binding changes, run cargo check -p fabric-python --locked.

Use snake_case for functions and variables; use PascalCase for Rust types and Python classes.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

If Python code or a Python-facing adapter changes, run just test-python.

In Python SDK, adapters, examples, and tests, follow the existing style, use type annotations for public APIs, and keep native binding declarations synchronized with their Rust implementations.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

**/*.{rs,py,pyi}: If public configuration types change, confirm schema snapshot tests in just test-rust pass and review generated schema diffs.
For schema or public contract changes, run both language suites and review changes under schemas/ and generated API references.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{py,pyi,rs}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

For Python SDK or PyO3 binding changes, use python-tests, run focused pytest tests first, then just test-python; rebuild with just build-python when native code or packaging changes.

Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py,toml}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

When editing version helpers, verify every nemo-fabric-* workspace package through Cargo metadata and reject a static version in python/pyproject.toml.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{toml,rs,py}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

Avoid blind repository-wide replacement of version-like strings; distinguish package-version references from examples and unrelated dependency versions.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

adapters/*/src/**/*.py

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

adapters/*/src/**/*.py: Implement adapters using the existing Fabric python or process runner and normalized request/result contracts; do not add a runner or one-off abstraction for a single adapter.
Treat normalized config, Fabric-resolved plans, and runtime_context as authoritative; reserve harness.settings for adapter-wide behavior and apply precedence in the order: normalized config, plans/context, harness settings, descriptor/default values.
Reject conflicting duplicate declarations and unsupported behavior with actionable errors naming the field and supported alternatives; never silently drop configuration.
Validate dependency versions, hooks, and credentials before harness invocation, and never expose credential values in outputs, errors, events, logs, or fixtures.
Forward only required system, selected credential, telemetry, and documented harness-specific environment variables; never forward or log unrelated environment values.
Maintain one local adapter host per Fabric runtime across ordered start → invoke* → stop operations. Emit one JSON lifecycle response per request on stdout and diagnostics on stderr.
Return harness-level invocation failures as successful lifecycle responses containing response: null, failed: true, and structured error fields (code, message, retryable, and optional metadata).
Do not emit NeMo Relay stream records on adapter stdout; return exactly one terminal lifecycle response while streaming occurs through the SDK-owned out-of-band endpoint.
Scope workspace, generated configuration, state, sessions, and artifacts to the resolved runtime context, and isolate stateful adapter instances by Fabric runtime ID.
Use start to initialize adapter-owned harness state, retain it for continuation across repeated invoke calls on the same runtime, and release it in stop.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

tests/adapters/**/*.py

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

tests/adapters/**/*.py: If an adapter or integration changes, run its focused tests.
For adapter behavior changes, run focused adapter tests under tests/adapters, then run just test-python.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.{rs,py}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

When adding functionality, include tests in the corresponding Rust crate or the relevant area under tests/.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.py

📄 CodeRabbit inference engine (.agents/skills/python-tests/SKILL.md)

tests/**/*.py: Use pytest to run Python tests.
Do not add @pytest.mark.asyncio to tests; async tests are automatically detected by the async runner.
Do not add -> None return type annotations to test functions.
When mocking a class, use unittest.mock.MagicMock or AsyncMock, using the spec argument when necessary, rather than defining a new class.
Prefix mocked class names with mock, not fake.
Prefer pytest fixtures over helper methods.
If a fixture is needed in multiple test files, define it once in conftest.py rather than repeating it.
Define fixtures using @pytest.fixture(name="<fixture_name>"[, scope="<scope>"]) and a <fixture_name>_fixture function; specify scope only when it is not function.
Prefer pytest.mark.parametrize over separate tests for different input types.
Use @pytest.mark.usefixtures when a fixture is needed but its returned value is unused or it returns no value.
Avoid defensive programming in tests; access expected values directly so missing data raises a clear failure, such as using results["data"] instead of results.get("data").
When adapter installation metadata changes, packaging metadata tests must directly assert that the root project depends unconditionally on the exact-version nemo-fabric-runtime distribution.
Packaging metadata tests must verify that each root harness extra delegates to the matching version of the leaf adapter's harness extra.
Packaging metadata tests must verify that bare leaf dependencies remain adapter-owned and that the root adapter-tests dependency group installs each leaf through its harness extra.
Packaging metadata tests must verify that every leaf provides full; only adapters importing NeMo Relay Python APIs provide relay, while adapters using an external Relay executable have full equal to harness.

Files:

• tests/adapters/test_deepagents.py

{tests/**,python/tests/**}

⚙️ CodeRabbit configuration file

{tests/**,python/tests/**}: Tests should cover the behavior promised by the changed API surface, including error paths, lifecycle cleanup, and SDK/native parity where relevant.

Files:

• tests/adapters/test_deepagents.py

🧠 Learnings (1)

📚 Learning: 2026-07-09T22:28:51.689Z

Learnt from: AjayThorve
Repo: NVIDIA/NeMo-Fabric PR: 43
File: adapters/claude-sdk/src/nemo_fabric_adapters/claude_sdk/adapter.py:164-168
Timestamp: 2026-07-09T22:28:51.689Z
Learning: In the NeMo-Fabric adapters, treat path values used in Fabric adapter configuration (including logic like `_resolve_path` in adapter.py) as config-root-relative. Do not apply `Path.expanduser()` (or otherwise apply `~`/home or shell-style expansion), because it will make the resolved paths normalize inconsistently across adapters. Also, do not rely on or add any resolution behavior that uses `harness.settings.cwd` as an override point for these adapter paths—`harness.settings.cwd` is explicitly unsupported in this adapter context.

Applied to files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

🪛 Ruff (0.16.1)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

[warning] 622-622: Do not catch blind exception: Exception

(BLE001)

---

[warning] 646-646: Do not catch blind exception: Exception

(BLE001)

---

[warning] 676-676: Do not catch blind exception: Exception

(BLE001)

tests/adapters/test_deepagents.py

[warning] 416-416: Unused function argument: fake_sdks

(ARG001)

---

[warning] 416-416: Unused function argument: fake_relay

(ARG001)

---

[warning] 428-428: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 428-428: Unused function argument: name

(ARG001)

---

[warning] 428-428: Unused function argument: scope_type

(ARG001)

---

[warning] 428-428: Unused function argument: kwargs

(ARG001)

---

[warning] 430-432: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 451-451: Unused function argument: fake_sdks

(ARG001)

---

[warning] 451-451: Unused function argument: fake_relay

(ARG001)

---

[warning] 462-462: Missing return type annotation for private function exploding_plugin

(ANN202)

---

[warning] 462-462: Unused function argument: config

(ARG001)

---

[warning] 464-464: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 479-479: Unused function argument: fake_sdks

(ARG001)

---

[warning] 479-479: Unused function argument: fake_relay

(ARG001)

---

[warning] 488-488: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 502-502: Unused function argument: fake_sdks

(ARG001)

---

[warning] 502-502: Unused function argument: fake_relay

(ARG001)

---

[warning] 509-509: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 509-509: Unused function argument: name

(ARG001)

---

[warning] 509-509: Unused function argument: scope_type

(ARG001)

---

[warning] 509-509: Unused function argument: kwargs

(ARG001)

---

[warning] 511-511: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 514-514: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 528-528: Unused function argument: fake_sdks

(ARG001)

---

[warning] 528-528: Unused function argument: fake_relay

(ARG001)

---

[warning] 535-535: Missing return type annotation for private function failing_scope

(ANN202)

---

[warning] 535-535: Unused function argument: name

(ARG001)

---

[warning] 535-535: Unused function argument: scope_type

(ARG001)

---

[warning] 535-535: Unused function argument: kwargs

(ARG001)

---

[warning] 536-536: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 556-556: Unused function argument: fake_sdks

(ARG001)

---

[warning] 556-556: Unused function argument: fake_relay

(ARG001)

---

[warning] 560-560: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 561-561: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 576-576: Unused function argument: fake_sdks

(ARG001)

---

[warning] 576-576: Unused function argument: fake_relay

(ARG001)

---

[warning] 582-582: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 583-583: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 586-586: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 586-586: Unused function argument: name

(ARG001)

---

[warning] 586-586: Unused function argument: scope_type

(ARG001)

---

[warning] 586-586: Unused function argument: kwargs

(ARG001)

---

[warning] 588-588: Avoid specifying long messages outside the exception class

(TRY003)

🔇 Additional comments (2)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py (2)

609-612: Move callback handler construction into the lifecycle try block.

self._callback_handler_type() can raise before Line 611. That bypasses the required invocation failure response for telemetry setup failures. Add a regression test for a failing callback handler constructor.

---

790-791: Do Not Return Raw Exception Text.

_error_text() copies exception messages into both invocation and telemetry output. Exceptions can contain credential or plugin configuration values. Return an allowlisted code or fixed safe message, and add a redaction regression test.

As per coding guidelines, “never expose credential values in outputs, errors, events, logs, or fixtures.”

Source: Coding guidelines

[SandyChapman]

The 41.38% is measured over the whole file rather than the diff; 5 of the 6 functions this PR adds have docstrings. Raising the full module to 80% would be a large unrelated diff, but happy to do it as a follow-up.

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py`:
- Around line 574-577: The invocation branch currently treats native and Relay
telemetry identically, causing native-only runs to depend on Relay. Update the
adapter’s startup and invocation flow around _configure_observability(),
_invoke_with_telemetry(), and self._telemetry_provider so Relay setup and
telemetry are selected only for provider "relay", while native configurations
use the native path; add a regression test that exercises native telemetry
without nemo_relay installed.

In `@tests/adapters/test_deepagents.py`:
- Around line 510-532: Extend
test_callback_handler_construction_failure_is_a_normalized_failure to stub
adapter.invoke_compiled_agent with an AsyncMock and assert it is not awaited
when callback construction fails. Also assert the response telemetry marks
degraded as true and telemetry.error contains the callback construction failure,
while preserving the existing normalized invocation error assertions.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 0645c070-c0be-4700-b1c2-afdde7675fc0

📥 Commits

Reviewing files that changed from the base of the PR and between b33a174 and 324c53c.

📒 Files selected for processing (3)

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

📜 Review details

⏰ Context from checks skipped due to timeout. (17)

• GitHub Check: Test (Python 3.14, macos-arm64)
• GitHub Check: Test (Python 3.12, windows-amd64)
• GitHub Check: Test (Python 3.13, windows-amd64)
• GitHub Check: Test (Python 3.14, linux-amd64)
• GitHub Check: Test (Python 3.12, linux-amd64)
• GitHub Check: Test (Python 3.14, windows-amd64)
• GitHub Check: Test (Python 3.11, macos-arm64)
• GitHub Check: Test (Python 3.13, macos-arm64)
• GitHub Check: Test (Python 3.12, linux-arm64)
• GitHub Check: Test (Python 3.13, linux-arm64)
• GitHub Check: Test (Python 3.11, windows-amd64)
• GitHub Check: Test (Python 3.14, linux-arm64)
• GitHub Check: Test (Python 3.11, linux-amd64)
• GitHub Check: Test (Python 3.12, macos-arm64)
• GitHub Check: Test (Python 3.11, linux-arm64)
• GitHub Check: Test (Python 3.13, linux-amd64)
• GitHub Check: Pre-commit

🧰 Additional context used

📓 Path-based instructions (24)

**/*.{md,rst}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Update documentation and examples in the same branch as the public API change.

Files:

• adapters/deepagents/README.md

**/*

📄 CodeRabbit inference engine (.agents/skills/karpathy-guidelines/SKILL.md)

**/*: Before implementing, explicitly state assumptions, surface ambiguity and tradeoffs, present multiple interpretations when relevant, and ask for clarification rather than silently deciding or proceeding when requirements are unclear.
Prefer the minimum code needed to solve the requested problem: avoid speculative features, unnecessary abstractions, unrequested flexibility, and handling of impossible scenarios; simplify overcomplicated solutions.
When editing existing code, make surgical changes only: do not modify unrelated code, comments, formatting, or pre-existing dead code; match the existing style, and remove only unused imports, variables, or functions introduced by your changes.
Define verifiable success criteria for each task, such as writing regression tests for bugs and invalid-input tests for validation, then verify the implementation against those criteria. For multi-step work, state a brief plan with a verification check for each step.

**/*: Always spell NVIDIA in all caps; do not use Nvidia, nvidia, nVidia, nVIDIA, or NV.
Use an NVIDIA before a noun, because the name begins with an “en” sound.
Do not add a registered trademark symbol after NVIDIA when referring to the company; use trademark symbols with product names only when required by the document type or legal guidance.
Verify official capitalization, spacing, hyphenation, and spelling for NVIDIA and third-party product names; do not rewrite official product names for grammar or title-case rules.
Precede NVIDIA product names with NVIDIA on first mention when natural and accurate, and link the first mention when the destination helps the reader.
On first use, include the company name and full model qualifier when it helps identify the model; preserve official capitalization and punctuation, and use shorter family names only after establishing the full name.
For learning-oriented and developer content, do not force trademark symbols unless explicitly required; for press, ...

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{md,mdx,rst}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-guide.md)

**/*.{md,mdx,rst}: For NeMo Fabric documentation, verify technical claims against the current repository, public API, or documented command before reviewing style.
Always spell NVIDIA in all caps; do not use Nvidia, nvidia, or NV.
Format commands, code elements, expressions, package names, file names, and paths as inline code.
Use descriptive link text; avoid raw URLs and weak anchors such as here or read more.
Use title case consistently for technical documentation headings.
Introduce code blocks, lists, tables, and images with complete sentences.
Write procedures as imperative, parallel steps; split long procedures into smaller tasks.
Prefer active voice, present tense, short sentences, contractions, and plain English while preserving necessary technical precision.
Use can for possibility and reserve may for permission.
Use after for temporal relationships instead of once, and prefer refer to over see when directing readers to another resource.
Avoid culture-specific idioms, unnecessary Latinisms, jokes, and marketing exaggeration in technical documentation.
Spell out months in body text, avoid ordinal dates, and use clear time zones.
Spell out whole numbers from zero through nine unless they are technical values, parameters, versions, or UI values; use numerals for 10 or greater and commas in thousands.
Do not add trademark symbols to learning-oriented documentation unless the source, platform, or legal guidance explicitly requires them.
Do not replace precise technical terms with simpler words when doing so would lose precision.
Do not flag passive voice when the actor is unknown or the action is the important part.
Do not rewrite API names, package names, command flags, or code literals for style.

**/*.{md,mdx,rst}: Use consistent title case for technical-document headings and table headers; avoid quotation marks, ampersands, and exclamation marks in headings, while preserving official product, event, research, and whitepaper title ...

Files:

• adapters/deepagents/README.md

**/*.{md,rst,txt,adoc}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/assets/nvidia-style-language-mechanics.md)

**/*.{md,rst,txt,adoc}: For technical documentation, use professional, active, conversational, engaging, precise, and plain-English prose. Prefer active voice, present tense, short sentences, and scannable paragraphs. Avoid casual or imprecise language, swearing, threats, insults, jokes, puns, culture-specific idioms, marketing exaggeration, and unsupported third-party comparisons.
Use can for possibility and reserve may for permission; use after for temporal order; use refer to for cross-references; prefer short direct sentences and specific verbs; avoid unnecessary please in technical documentation.
Prefer active voice when the actor matters. Passive voice is acceptable when the actor is unknown or irrelevant, when the action or result is the focus, or in programmer documentation.
Use natural contractions in conversational technical prose, but do not force them in formal legal copy, API references, or generated text.
Prefer simpler English over Latinisms: use for example or such as instead of e.g., and so on instead of etc., that is instead of i.e., compared to instead of vs., and by, through, or using instead of via. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Use that without commas for essential clauses, and which with commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such as June 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space before a.m. or p.m.; use ET and PT for needed time zones; avoid 24/7; and prefer from 12:30 to 1:00 p.m. for prose ranges.
Format numbers consistently: spell out zero through nine in body text, use numerals for 10 or greater and for technical values, use commas in thousands, do not begin a sentence with a numeral, spell out ordinals, and use numerals consistently within a category wh...

Files:

• adapters/deepagents/README.md

**/*.{md,mdx,yml,py,rs,sh}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

Keep documentation aligned with current NeMo Fabric behavior, repository layout, entry points, commands, package names, APIs, bindings, and support claims.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{md,mdx}

📄 CodeRabbit inference engine (.agents/skills/review-doc-style/SKILL.md)

**/*.{md,mdx}: Use the full product name NVIDIA NeMo Fabric on first use, typically in the title and H1; use NeMo Fabric thereafter. Use fabric alone only for the CLI tool and surround it with backticks.
Treat incorrect or stale commands, package names, paths, APIs, support claims, procedures, examples, terminology, or public behavior documentation as blocking issues.
Capitalize NVIDIA correctly and format code, commands, paths, and filenames as inline code where needed.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences; ensure examples match current APIs and build commands.
Use descriptive anchor text, avoid raw URLs and generic labels such as here, and use repository-relative .mdx paths for links within docs/.
Prefer active voice, present tense, short sentences, plain English, consistent terminology, and imperative, parallel, scannable procedures.
Use after instead of once when expressing temporal sequence, and use can rather than may when describing possibility rather than permission.
Avoid ambiguous numeric dates and ordinal dates in body text.
For learning-oriented documentation, do not force trademark symbols unless the source document explicitly requires them.
When reporting documentation-review findings, lead with Must fix, Should fix, and Nice to have categories; include file path, line reference, current problem, rationale, and a concrete rewrite or direction.

Files:

• adapters/deepagents/README.md

**/*.md

📄 CodeRabbit inference engine (CONTRIBUTING.md)

Update relevant SDK, API reference, adapter, example, integration, and support documentation when the corresponding public surface changes.

Files:

• adapters/deepagents/README.md

**/*.{html,md}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

HTML and Markdown files must use the specified SPDX HTML-comment header.

Files:

• adapters/deepagents/README.md

adapters/*/{README.md,fabric-adapter.json}

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Document installation, supported configuration, harness-only settings, credentials, lifecycle, telemetry, artifacts, limitations, and focused test commands; keep documentation consistent with descriptor claims.

Files:

• adapters/deepagents/README.md

adapters/*/README.md

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

Provide a credential-free fixture covering plan, doctor, and run, plus a canonical typed SDK example and, when required, a canonical harness-native YAML fixture.

Files:

• adapters/deepagents/README.md

{adapters/**,examples/**}

⚙️ CodeRabbit configuration file

{adapters/**,examples/**}: Review adapter and example changes for command correctness, config/schema consistency, artifact handling, and compatibility with the public NeMo Fabric contracts.

Files:

• adapters/deepagents/README.md
• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

{*.md,**/*.md,**/*.mdx,**/*.ipynb}

⚙️ CodeRabbit configuration file

{*.md,**/*.md,**/*.mdx,**/*.ipynb}: Enforce the product name in user-facing prose: use "NVIDIA NeMo Fabric" on first use and "NeMo Fabric" thereafter. Flag standalone capitalized "Fabric" when it refers to the product. Do not flag the lowercase fabric CLI command, package/import/crate names, code identifiers, API symbols, configuration keys, file paths, or unrelated generic uses of the word.

Files:

• adapters/deepagents/README.md

**/*.{rs,py,pyi,json,yaml,yml}

📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)

Determine and update every affected public surface, including the CLI, PyO3 bindings, Python SDK, type stubs, schemas, and adapter contract, so they remain in parity.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py}

📄 CodeRabbit inference engine (.agents/skills/prepare-pr/SKILL.md)

For native binding changes, run cargo check -p fabric-python --locked.

Use snake_case for functions and variables; use PascalCase for Rust types and Python classes.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

If Python code or a Python-facing adapter changes, run just test-python.

In Python SDK, adapters, examples, and tests, follow the existing style, use type annotations for public APIs, and keep native binding declarations synchronized with their Rust implementations.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py,pyi}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

**/*.{rs,py,pyi}: If public configuration types change, confirm schema snapshot tests in just test-rust pass and review generated schema diffs.
For schema or public contract changes, run both language suites and review changes under schemas/ and generated API references.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{py,pyi,rs}

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

For Python SDK or PyO3 binding changes, use python-tests, run focused pytest tests first, then just test-python; rebuild with just build-python when native code or packaging changes.

Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{rs,py,toml}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

When editing version helpers, verify every nemo-fabric-* workspace package through Cargo metadata and reject a static version in python/pyproject.toml.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

**/*.{toml,rs,py}

📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)

Avoid blind repository-wide replacement of version-like strings; distinguish package-version references from examples and unrelated dependency versions.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py
• tests/adapters/test_deepagents.py

adapters/*/src/**/*.py

📄 CodeRabbit inference engine (.agents/skills/contribute-adapter/SKILL.md)

adapters/*/src/**/*.py: Implement adapters using the existing Fabric python or process runner and normalized request/result contracts; do not add a runner or one-off abstraction for a single adapter.
Treat normalized config, Fabric-resolved plans, and runtime_context as authoritative; reserve harness.settings for adapter-wide behavior and apply precedence in the order: normalized config, plans/context, harness settings, descriptor/default values.
Reject conflicting duplicate declarations and unsupported behavior with actionable errors naming the field and supported alternatives; never silently drop configuration.
Validate dependency versions, hooks, and credentials before harness invocation, and never expose credential values in outputs, errors, events, logs, or fixtures.
Forward only required system, selected credential, telemetry, and documented harness-specific environment variables; never forward or log unrelated environment values.
Maintain one local adapter host per Fabric runtime across ordered start → invoke* → stop operations. Emit one JSON lifecycle response per request on stdout and diagnostics on stderr.
Return harness-level invocation failures as successful lifecycle responses containing response: null, failed: true, and structured error fields (code, message, retryable, and optional metadata).
Do not emit NeMo Relay stream records on adapter stdout; return exactly one terminal lifecycle response while streaming occurs through the SDK-owned out-of-band endpoint.
Scope workspace, generated configuration, state, sessions, and artifacts to the resolved runtime context, and isolate stateful adapter instances by Fabric runtime ID.
Use start to initialize adapter-owned harness state, retain it for continuation across repeated invoke calls on the same runtime, and release it in stop.

Files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

tests/adapters/**/*.py

📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)

tests/adapters/**/*.py: If an adapter or integration changes, run its focused tests.
For adapter behavior changes, run focused adapter tests under tests/adapters, then run just test-python.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.{rs,py}

📄 CodeRabbit inference engine (CONTRIBUTING.md)

When adding functionality, include tests in the corresponding Rust crate or the relevant area under tests/.

Files:

• tests/adapters/test_deepagents.py

tests/**/*.py

📄 CodeRabbit inference engine (.agents/skills/python-tests/SKILL.md)

tests/**/*.py: Use pytest to run Python tests.
Do not add @pytest.mark.asyncio to tests; async tests are automatically detected by the async runner.
Do not add -> None return type annotations to test functions.
When mocking a class, use unittest.mock.MagicMock or AsyncMock, using the spec argument when necessary, rather than defining a new class.
Prefix mocked class names with mock, not fake.
Prefer pytest fixtures over helper methods.
If a fixture is needed in multiple test files, define it once in conftest.py rather than repeating it.
Define fixtures using @pytest.fixture(name="<fixture_name>"[, scope="<scope>"]) and a <fixture_name>_fixture function; specify scope only when it is not function.
Prefer pytest.mark.parametrize over separate tests for different input types.
Use @pytest.mark.usefixtures when a fixture is needed but its returned value is unused or it returns no value.
Avoid defensive programming in tests; access expected values directly so missing data raises a clear failure, such as using results["data"] instead of results.get("data").
When adapter installation metadata changes, packaging metadata tests must directly assert that the root project depends unconditionally on the exact-version nemo-fabric-runtime distribution.
Packaging metadata tests must verify that each root harness extra delegates to the matching version of the leaf adapter's harness extra.
Packaging metadata tests must verify that bare leaf dependencies remain adapter-owned and that the root adapter-tests dependency group installs each leaf through its harness extra.
Packaging metadata tests must verify that every leaf provides full; only adapters importing NeMo Relay Python APIs provide relay, while adapters using an external Relay executable have full equal to harness.

Files:

• tests/adapters/test_deepagents.py

{tests/**,python/tests/**}

⚙️ CodeRabbit configuration file

{tests/**,python/tests/**}: Tests should cover the behavior promised by the changed API surface, including error paths, lifecycle cleanup, and SDK/native parity where relevant.

Files:

• tests/adapters/test_deepagents.py

🧠 Learnings (1)

📚 Learning: 2026-07-09T22:28:51.689Z

Learnt from: AjayThorve
Repo: NVIDIA/NeMo-Fabric PR: 43
File: adapters/claude-sdk/src/nemo_fabric_adapters/claude_sdk/adapter.py:164-168
Timestamp: 2026-07-09T22:28:51.689Z
Learning: In the NeMo-Fabric adapters, treat path values used in Fabric adapter configuration (including logic like `_resolve_path` in adapter.py) as config-root-relative. Do not apply `Path.expanduser()` (or otherwise apply `~`/home or shell-style expansion), because it will make the resolved paths normalize inconsistently across adapters. Also, do not rely on or add any resolution behavior that uses `harness.settings.cwd` as an override point for these adapter paths—`harness.settings.cwd` is explicitly unsupported in this adapter context.

Applied to files:

• adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

🪛 Ruff (0.16.1)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py

[warning] 627-627: Do not catch blind exception: Exception

(BLE001)

---

[warning] 629-629: Do not catch blind exception: Exception

(BLE001)

---

[warning] 653-653: Do not catch blind exception: Exception

(BLE001)

---

[warning] 683-683: Do not catch blind exception: Exception

(BLE001)

tests/adapters/test_deepagents.py

[warning] 416-416: Unused function argument: fake_sdks

(ARG001)

---

[warning] 416-416: Unused function argument: fake_relay

(ARG001)

---

[warning] 428-428: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 428-428: Unused function argument: name

(ARG001)

---

[warning] 428-428: Unused function argument: scope_type

(ARG001)

---

[warning] 428-428: Unused function argument: kwargs

(ARG001)

---

[warning] 430-432: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 451-451: Unused function argument: fake_sdks

(ARG001)

---

[warning] 451-451: Unused function argument: fake_relay

(ARG001)

---

[warning] 462-462: Missing return type annotation for private function exploding_plugin

(ANN202)

---

[warning] 462-462: Unused function argument: config

(ARG001)

---

[warning] 464-464: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 479-479: Unused function argument: fake_sdks

(ARG001)

---

[warning] 479-479: Unused function argument: fake_relay

(ARG001)

---

[warning] 489-489: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 489-489: Unused function argument: name

(ARG001)

---

[warning] 489-489: Unused function argument: scope_type

(ARG001)

---

[warning] 489-489: Unused function argument: kwargs

(ARG001)

---

[warning] 491-491: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 494-494: Missing return type annotation for private function exploding_plugin

(ANN202)

---

[warning] 494-494: Unused function argument: config

(ARG001)

---

[warning] 496-496: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 511-511: Unused function argument: fake_sdks

(ARG001)

---

[warning] 511-511: Unused function argument: fake_relay

(ARG001)

---

[warning] 519-519: Missing return type annotation for private function exploding_handler

Add return type annotation: Never

(ANN202)

---

[warning] 520-520: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 536-536: Unused function argument: fake_sdks

(ARG001)

---

[warning] 536-536: Unused function argument: fake_relay

(ARG001)

---

[warning] 545-545: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 559-559: Unused function argument: fake_sdks

(ARG001)

---

[warning] 559-559: Unused function argument: fake_relay

(ARG001)

---

[warning] 566-566: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 566-566: Unused function argument: name

(ARG001)

---

[warning] 566-566: Unused function argument: scope_type

(ARG001)

---

[warning] 566-566: Unused function argument: kwargs

(ARG001)

---

[warning] 568-568: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 571-571: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 585-585: Unused function argument: fake_sdks

(ARG001)

---

[warning] 585-585: Unused function argument: fake_relay

(ARG001)

---

[warning] 592-592: Missing return type annotation for private function failing_scope

(ANN202)

---

[warning] 592-592: Unused function argument: name

(ARG001)

---

[warning] 592-592: Unused function argument: scope_type

(ARG001)

---

[warning] 592-592: Unused function argument: kwargs

(ARG001)

---

[warning] 593-593: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 613-613: Unused function argument: fake_sdks

(ARG001)

---

[warning] 613-613: Unused function argument: fake_relay

(ARG001)

---

[warning] 617-617: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 618-618: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 633-633: Unused function argument: fake_sdks

(ARG001)

---

[warning] 633-633: Unused function argument: fake_relay

(ARG001)

---

[warning] 639-639: Missing return type annotation for private function boom

Add return type annotation: Never

(ANN202)

---

[warning] 640-640: Avoid specifying long messages outside the exception class

(TRY003)

---

[warning] 643-643: Missing return type annotation for private function exploding_scope

(ANN202)

---

[warning] 643-643: Unused function argument: name

(ARG001)

---

[warning] 643-643: Unused function argument: scope_type

(ARG001)

---

[warning] 643-643: Unused function argument: kwargs

(ARG001)

---

[warning] 645-645: Avoid specifying long messages outside the exception class

(TRY003)

🔇 Additional comments (4)

adapters/deepagents/src/nemo_fabric_adapters/deepagents/adapter.py (1)

579-596: LGTM!

Also applies to: 598-637, 639-685, 785-809, 862-896

tests/adapters/test_deepagents.py (2)

389-508: LGTM!

---

535-658: LGTM!

adapters/deepagents/README.md (1)

211-227: LGTM!

[AjayThorve]

The agent-outcome/telemetry-outcome split is the right containment boundary, and the real-agent runs confirm that response and workspace effects survive the teardown fault. I am requesting changes only for cross-invocation containment: the first degraded turn currently leaves Relay state corrupted while a later turn can be reported clean.

The concurrency/root fix likely belongs in NeMo Relay, not in this adapter. Current Relay 0.7.x has task-context forking primitives from NVIDIA/NeMo-Relay#589, but its LangChain callback does not apply them at overlapping sibling-run boundaries. This Fabric PR should establish a safe post-fault invariant and link/track the Relay-side fix rather than reimplementing Relay internals here.

The full NVBug acceptance still requires the Relay fix, complete ATOF/ATIF without scope-pop errors, and the Platform test matrices. Once the persistent-runtime invariant below is addressed, I would be comfortable merging this as containment while the Relay fix proceeds.

[AjayThorve]

Validated the current head: the persistent-runtime containment now keeps poisoned runtimes degraded across ordered invokes and stop/start, and the real Relay callback regression covers the A/B/A-end/B-end leak. Focused validation passed 60 tests; just test-python passed 730 tests with 16 skipped; hosted checks are green.

Approving this as the Fabric containment fix. The root scope-stack repair remains owned by NeMo Relay. I left one non-blocking P2 inline for the faulting-turn artifact contract.

[AjayThorve]

/merge