feat: add normalized MCP and skills to Codex - #88
Conversation
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
|
Note Reviews pausedIt 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 Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
WalkthroughThe Codex adapter now accepts normalized MCP and skills configuration, validates native servers and skill directories, registers skill roots through the SDK, and injects MCP servers into thread configuration. Documentation, dependency metadata, and tests cover the updated runtime and capability behavior. ChangesCodex Native Capabilities
Estimated code review effort: 4 (Complex) | ~45 minutes Sequence Diagram(s)sequenceDiagram
participant FabricConfig
participant CodexAdapter
participant CodexSDK
participant CodexThread
FabricConfig->>CodexAdapter: normalized MCP servers and skill paths
CodexAdapter->>CodexAdapter: validate transports and resolve SKILL.md directories
CodexAdapter->>CodexSDK: register skill roots with extraRoots
CodexAdapter->>CodexThread: send prompt with mcp_servers configuration
CodexThread-->>CodexAdapter: thread turn result
Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
|
📖 Fern docs preview: https://nvidia-preview-pull-request-88.docs.buildwithfern.com/nemo/fabric |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 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/codex/src/nemo_fabric_adapters/codex/adapter.py`:
- Around line 241-249: Update _native_skill_paths to default skill_paths to an
empty list only when the capability key is absent; preserve falsy values such as
None, an empty string, an empty mapping, or False for validation so they raise
AdapterConfigError instead of disabling skills. Add a parameterized regression
test covering these falsy non-list values.
In `@docs/integrations/codex.mdx`:
- Around line 33-38: Update the adapter description to state that each skill
directory is resolved, validated by checking for its SKILL.md file, and then
registered as a process-scoped Codex skill root; do not say the directory is
resolved to the file. Preserve the existing statements about skill discovery and
startup validation.
In `@tests/adapters/test_codex_adapter.py`:
- Line 117: Add a failure-path test around the private skill-registration seam
in the Codex adapter, configuring the mocked client without a usable
`_client.request`. Assert the adapter returns `codex_invalid_configuration`,
does not start a thread, and closes the SDK client during cleanup; preserve the
existing success-path setup and assertions.
- Around line 199-202: Replace the monkeypatch.setenv call in
test_sdk_maps_native_mcp_servers_into_thread_config with an os.environ
assignment for FABRIC_TEST_MCP_URL, preserving the same test value and
environment setup.
🪄 Autofix (Beta)
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: e8be5fd1-e640-4450-aaf5-6e02194a6672
📒 Files selected for processing (5)
adapters/codex/README.mdadapters/codex/fabric-adapter.jsonadapters/codex/src/nemo_fabric_adapters/codex/adapter.pydocs/integrations/codex.mdxtests/adapters/test_codex_adapter.py
📜 Review details
⏰ Context from checks skipped due to timeout. (11)
- GitHub Check: Build and publish docs
- GitHub Check: Test (Python 3.12, arm64)
- GitHub Check: Test (Python 3.11, arm64)
- GitHub Check: Test (Python 3.14, arm64)
- GitHub Check: Test (Python 3.13, arm64)
- GitHub Check: Test (Python 3.14, x86_64)
- GitHub Check: Test (Python 3.13, x86_64)
- GitHub Check: Test (Python 3.11, x86_64)
- GitHub Check: Test (Python 3.12, x86_64)
- GitHub Check: Build wheels (x86_64)
- GitHub Check: Pre-commit
🧰 Additional context used
📓 Path-based instructions (26)
**/*.{json,jsonc}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Public contract changes must keep checked-in JSON Schema snapshots synchronized.
Files:
adapters/codex/fabric-adapter.json
**/*
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*: All source files must include the specified SPDX copyright and Apache-2.0 license header using the comment syntax appropriate to the file type.
Release tags must use raw Rust-compatible SemVer without a leadingv, such as0.1.0or0.1.0-rc.1.
**/*: 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.
**/*: Keep pull request branch scope coherent and reviewable.
Run relevant tests undervalidate-changebefore opening or updating a pull request.
Format changed files with the language-native formatter.
Update documentation and examples for public behavior changes.
Update dependent maintainer or consumer guidance when code changes affect APIs, bindings, commands, paths, packaging guidance, or best practices.
Use Conventional Commit style for pull request titles:<type>: <concise imperative summary>, choosing the type from the actual change surface. Usefixonly for user-facing or runtime product-code bug fixes.
A pull request body must include#### Overview,#### Details,#### Validation,#### Where should the reviewer start?, and `#### Related ...
Files:
adapters/codex/fabric-adapter.jsondocs/integrations/codex.mdxtests/adapters/test_codex_adapter.pyadapters/codex/src/nemo_fabric_adapters/codex/adapter.pyadapters/codex/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/codex/fabric-adapter.jsontests/adapters/test_codex_adapter.pyadapters/codex/src/nemo_fabric_adapters/codex/adapter.py
{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 Fabric contracts.
Files:
adapters/codex/fabric-adapter.jsonadapters/codex/src/nemo_fabric_adapters/codex/adapter.pyadapters/codex/README.md
**/*.{md,mdx,html}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Changes affecting public behavior, adapters, examples, or workspace structure must update the corresponding documentation; public API changes require updated SDK or API reference documentation.
Files:
docs/integrations/codex.mdxadapters/codex/README.md
**/*.{md,mdx}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
For docs site changes, run
just docsto regenerate Python and Rust API references and validate Fern configuration.
**/*.{md,mdx}: Prioritize factual accuracy in NeMo Fabric documentation and keep commands, package names, APIs, file paths, repository layout, entry points, support claims, examples, and procedures aligned with current repository behavior.
Update relevant entry-point documentation when public behavior changes, includingREADME.md,docs/index.yml, package or crate READMEs, and adapter or integration READMEs.
Use{/* ... */}delimiters for top-of-file SPDX comments in MDX files, not HTML comment delimiters.
CapitalizeNVIDIAcorrectly and use consistent current repository terminology, product names, APIs, and feature names.
Format commands, code, expressions, file names, paths, and filenames as inline code where appropriate.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences.
Use descriptive link text instead of raw URLs or generic labels such ashere.
Write procedures as short, imperative, parallel, easy-to-scan steps; prefer active voice, present tense, plain English, and concise sentences.
Useafterinstead ofoncewhen expressing temporal sequence, and usecaninstead ofmaywhen describing possibility rather than permission.
Use unambiguous date formats and avoid ordinal dates in body text.
When reviewing documentation, report findings in severity order underMust fix,Should fix, andNice to have, with file paths, line references, explanations, and concrete rewrites or directions.
Files:
docs/integrations/codex.mdxadapters/codex/README.md
**/*.mdx
📄 CodeRabbit inference engine (CONTRIBUTING.md)
MDX files must use the specified SPDX header in a JSX comment.
In MDX files, use JSX comment delimiters (
{/*and*/}) for top-of-file comments, including SPDX headers; do not use HTML comments.
Files:
docs/integrations/codex.mdx
{README.md,docs/**/*.{md,mdx,yml},examples/**/*.{md,mdx,yml}}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Keep package names, repository references, and build commands current in documentation and examples.
Files:
docs/integrations/codex.mdx
{docs/**/*.{md,mdx,yml},examples/**/*.{md,mdx,yml}}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Update relevant getting-started, reference, adapter, and example documentation when the corresponding examples or adapters change.
Files:
docs/integrations/codex.mdx
docs/**/*.{md,mdx,yml}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Run
just docswhen the documentation site changes.
Files:
docs/integrations/codex.mdx
{docs/**/*,.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml,justfile}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Use the current install, import, build, test, clean, and documentation commands consistently in documentation, examples, CI workflows, and just recipes.
Files:
docs/integrations/codex.mdx
{docs/**/*,.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Reflect public packaging changes in release-facing documentation and examples.
Files:
docs/integrations/codex.mdx
**/*.{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 spellNVIDIAin all caps; do not useNvidia,nvidia, orNV.
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 ashereorread 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.
Usecanfor possibility and reservemayfor permission.
Useafterfor temporal relationships instead ofonce, and preferrefer tooverseewhen 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:
docs/integrations/codex.mdxadapters/codex/README.md
docs/**/*
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
For documentation or examples changes, run
just docswhen practical and verify documented commands against the current repository.
Files:
docs/integrations/codex.mdx
{docs/**,README.md,AGENTS.md}
⚙️ CodeRabbit configuration file
{docs/**,README.md,AGENTS.md}: Review documentation for technical accuracy against the current API, command correctness, and consistency with generated schemas.
Files:
docs/integrations/codex.mdx
**/*.py
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.py: Python public APIs must use type annotations, and native Python binding declarations must remain synchronized with their Rust implementations.
Python files must begin with the specified#SPDX copyright and Apache-2.0 license header.
Files:
tests/adapters/test_codex_adapter.pyadapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{rs,py}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.{rs,py}: Usesnake_casefor Rust and Python functions and variables; usePascalCasefor Rust types and Python classes.
Run tests for every language surface affected by a change. Changes touching the Rust core or public schemas require both Rust and Python test suites.
Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.
Files:
tests/adapters/test_codex_adapter.pyadapters/codex/src/nemo_fabric_adapters/codex/adapter.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.asyncioto tests; async tests are automatically detected by the async runner.
Do not add-> Nonereturn annotations to test functions.
When mocking a class, useunittest.mock.MagicMockorAsyncMock, supplyingspecwhen necessary; do not define a new mock class.
Prefix mocked class names withmock, notfake.
Prefer pytest fixtures over helper methods.
Define shared fixtures inconftest.pyrather than repeating them across test files.
Define fixtures using@pytest.fixture(name="<fixture_name>"[, scope="<scope>"])and a<fixture_name>_fixturefunction; specifyscopeonly when it is notfunction.
Preferpytest.mark.parametrizeover separate tests for different input types.
Use@pytest.mark.usefixtureswhen a fixture is needed but its return value is unused.
Useos.environto modify environment variables in tests; do not usemonkeypatch.setenv, because the autouserestore_environ_fixtureintests/conftest.pyrestores the environment after each test.
Avoid defensive programming in tests; access expected data directly so missing data raises a clear error instead of being silently tolerated.
Run focused tests withuv run pytest -k "<pattern>"and all tests withuv run pytest.
Files:
tests/adapters/test_codex_adapter.py
**/*.{py,pyi}
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When Python code or a Python-facing adapter changes, run
just test-python.
Files:
tests/adapters/test_codex_adapter.pyadapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{rs,py,pyi,toml}
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When the PyO3 bridge or package metadata changes, run
just build-pythonandcargo check -p fabric-python --locked.
Files:
tests/adapters/test_codex_adapter.pyadapters/codex/src/nemo_fabric_adapters/codex/adapter.py
tests/adapters/**/*
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When an adapter or integration changes, run its focused tests under
tests/adapters, followed byjust test-python.
Files:
tests/adapters/test_codex_adapter.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_codex_adapter.py
**/README.md
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Update an adapter or example
README.mdwhen that adapter or example surface changes.
Files:
adapters/codex/README.md
**/*.{html,md}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
HTML and Markdown files must use the specified SPDX header in an HTML comment.
Files:
adapters/codex/README.md
**/*.{md,rst}
📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)
Update documentation and examples in the same branch as the public API change.
Verify README and documentation entry points, package names, paths, examples, and public commands remain current after changes.
Files:
adapters/codex/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.
Usecanfor possibility and reservemayfor permission; useafterfor temporal order; userefer tofor cross-references; prefer short direct sentences and specific verbs; avoid unnecessarypleasein 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: usefor exampleorsuch asinstead ofe.g.,and so oninstead ofetc.,that isinstead ofi.e.,compared toinstead ofvs., andby,through, orusinginstead ofvia. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Usethatwithout commas for essential clauses, andwhichwith commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such asJune 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space beforea.m.orp.m.; useETandPTfor needed time zones; avoid24/7; and preferfrom 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/codex/README.md
🧠 Learnings (2)
📚 Learning: 2026-06-29T22:34:52.407Z
Learnt from: AjayThorve
Repo: NVIDIA/NeMo-Fabric PR: 27
File: adapters/codex-cli/fabric-adapter.json:13-15
Timestamp: 2026-06-29T22:34:52.407Z
Learning: In NeMo-Fabric adapter manifest files (e.g., `*/fabric-adapter.json`), keep `config.accepts` limited to the top-level Fabric capability sections that `resolve_capability_plan` consumes (such as `models`, `tools`, `mcp`, `skills`, `telemetry`). Do not add adapter-owned `harness.settings` keys to `config.accepts`; `harness.settings` should remain adapter-owned and be passed through unchanged.
Applied to files:
adapters/codex/fabric-adapter.json
📚 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/codex/src/nemo_fabric_adapters/codex/adapter.py
🧬 Code graph analysis (1)
tests/adapters/test_codex_adapter.py (1)
python/src/nemo_fabric/types.py (2)
add_mcp_server(720-738)add_skill_path(740-744)
🔇 Additional comments (5)
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py (1)
13-13: LGTM!Also applies to: 29-29, 86-90, 180-240, 251-299, 647-649, 738-738, 896-904, 933-933
adapters/codex/fabric-adapter.json (1)
11-11: LGTM!adapters/codex/README.md (1)
82-95: LGTM!Also applies to: 104-105, 170-172
docs/integrations/codex.mdx (1)
2-3: LGTM!Also applies to: 12-31, 40-44
tests/adapters/test_codex_adapter.py (1)
9-9: LGTM!Also applies to: 195-198, 203-236, 264-291, 657-672, 695-728
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 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 `@docs/integrations/codex.mdx`:
- Around line 51-52: Update the codex_bin documentation to state that overrides
accept either an absolute path or a path relative to the Fabric config root.
Keep the existing intentional app-server override guidance and ensure the
wording matches the adapter’s supported path behavior.
🪄 Autofix (Beta)
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: d27963ab-8c1c-4f24-b114-de8d882e553c
⛔ Files ignored due to path filters (2)
adapters/codex/uv.lockis excluded by!**/*.lockuv.lockis excluded by!**/*.lock
📒 Files selected for processing (6)
ATTRIBUTIONS-Python.mdadapters/codex/README.mdadapters/codex/pyproject.tomladapters/codex/src/nemo_fabric_adapters/codex/adapter.pydocs/integrations/codex.mdxpyproject.toml
💤 Files with no reviewable changes (1)
- pyproject.toml
📜 Review details
⏰ Context from checks skipped due to timeout. (10)
- GitHub Check: Test (Python 3.14, arm64)
- GitHub Check: Test (Python 3.12, x86_64)
- GitHub Check: Test (Python 3.14, x86_64)
- GitHub Check: Test (Python 3.13, x86_64)
- GitHub Check: Test (Python 3.13, arm64)
- GitHub Check: Test (Python 3.11, arm64)
- GitHub Check: Test (Python 3.12, arm64)
- GitHub Check: Test (Python 3.11, x86_64)
- GitHub Check: Build and publish docs
- GitHub Check: Pre-commit
🧰 Additional context used
📓 Path-based instructions (25)
**/*.{md,mdx,html}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Changes affecting public behavior, adapters, examples, or workspace structure must update the corresponding documentation; public API changes require updated SDK or API reference documentation.
Files:
ATTRIBUTIONS-Python.mddocs/integrations/codex.mdxadapters/codex/README.md
**/*.{md,mdx}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
For docs site changes, run
just docsto regenerate Python and Rust API references and validate Fern configuration.
**/*.{md,mdx}: Prioritize factual accuracy in NeMo Fabric documentation and keep commands, package names, APIs, file paths, repository layout, entry points, support claims, examples, and procedures aligned with current repository behavior.
Update relevant entry-point documentation when public behavior changes, includingREADME.md,docs/index.yml, package or crate READMEs, and adapter or integration READMEs.
Use{/* ... */}delimiters for top-of-file SPDX comments in MDX files, not HTML comment delimiters.
CapitalizeNVIDIAcorrectly and use consistent current repository terminology, product names, APIs, and feature names.
Format commands, code, expressions, file names, paths, and filenames as inline code where appropriate.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences.
Use descriptive link text instead of raw URLs or generic labels such ashere.
Write procedures as short, imperative, parallel, easy-to-scan steps; prefer active voice, present tense, plain English, and concise sentences.
Useafterinstead ofoncewhen expressing temporal sequence, and usecaninstead ofmaywhen describing possibility rather than permission.
Use unambiguous date formats and avoid ordinal dates in body text.
When reviewing documentation, report findings in severity order underMust fix,Should fix, andNice to have, with file paths, line references, explanations, and concrete rewrites or directions.
Files:
ATTRIBUTIONS-Python.mddocs/integrations/codex.mdxadapters/codex/README.md
**/*
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*: All source files must include the specified SPDX copyright and Apache-2.0 license header using the comment syntax appropriate to the file type.
Release tags must use raw Rust-compatible SemVer without a leadingv, such as0.1.0or0.1.0-rc.1.
**/*: 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.
**/*: Keep pull request branch scope coherent and reviewable.
Run relevant tests undervalidate-changebefore opening or updating a pull request.
Format changed files with the language-native formatter.
Update documentation and examples for public behavior changes.
Update dependent maintainer or consumer guidance when code changes affect APIs, bindings, commands, paths, packaging guidance, or best practices.
Use Conventional Commit style for pull request titles:<type>: <concise imperative summary>, choosing the type from the actual change surface. Usefixonly for user-facing or runtime product-code bug fixes.
A pull request body must include#### Overview,#### Details,#### Validation,#### Where should the reviewer start?, and `#### Related ...
Files:
ATTRIBUTIONS-Python.mddocs/integrations/codex.mdxadapters/codex/README.mdadapters/codex/pyproject.tomladapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{html,md}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
HTML and Markdown files must use the specified SPDX header in an HTML comment.
Files:
ATTRIBUTIONS-Python.mdadapters/codex/README.md
**/*.{md,rst}
📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)
Update documentation and examples in the same branch as the public API change.
Verify README and documentation entry points, package names, paths, examples, and public commands remain current after changes.
Files:
ATTRIBUTIONS-Python.mdadapters/codex/README.md
**/*.{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 spellNVIDIAin all caps; do not useNvidia,nvidia, orNV.
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 ashereorread 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.
Usecanfor possibility and reservemayfor permission.
Useafterfor temporal relationships instead ofonce, and preferrefer tooverseewhen 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:
ATTRIBUTIONS-Python.mddocs/integrations/codex.mdxadapters/codex/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.
Usecanfor possibility and reservemayfor permission; useafterfor temporal order; userefer tofor cross-references; prefer short direct sentences and specific verbs; avoid unnecessarypleasein 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: usefor exampleorsuch asinstead ofe.g.,and so oninstead ofetc.,that isinstead ofi.e.,compared toinstead ofvs., andby,through, orusinginstead ofvia. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Usethatwithout commas for essential clauses, andwhichwith commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such asJune 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space beforea.m.orp.m.; useETandPTfor needed time zones; avoid24/7; and preferfrom 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:
ATTRIBUTIONS-Python.mdadapters/codex/README.md
**/*.mdx
📄 CodeRabbit inference engine (CONTRIBUTING.md)
MDX files must use the specified SPDX header in a JSX comment.
In MDX files, use JSX comment delimiters (
{/*and*/}) for top-of-file comments, including SPDX headers; do not use HTML comments.
Files:
docs/integrations/codex.mdx
{README.md,docs/**/*.{md,mdx,yml},examples/**/*.{md,mdx,yml}}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Keep package names, repository references, and build commands current in documentation and examples.
Files:
docs/integrations/codex.mdx
{docs/**/*.{md,mdx,yml},examples/**/*.{md,mdx,yml}}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Update relevant getting-started, reference, adapter, and example documentation when the corresponding examples or adapters change.
Files:
docs/integrations/codex.mdx
docs/**/*.{md,mdx,yml}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Run
just docswhen the documentation site changes.
Files:
docs/integrations/codex.mdx
{docs/**/*,.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml,justfile}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Use the current install, import, build, test, clean, and documentation commands consistently in documentation, examples, CI workflows, and just recipes.
Files:
docs/integrations/codex.mdx
{docs/**/*,.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Reflect public packaging changes in release-facing documentation and examples.
Files:
docs/integrations/codex.mdx
docs/**/*
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
For documentation or examples changes, run
just docswhen practical and verify documented commands against the current repository.
Files:
docs/integrations/codex.mdx
{docs/**,README.md,AGENTS.md}
⚙️ CodeRabbit configuration file
{docs/**,README.md,AGENTS.md}: Review documentation for technical accuracy against the current API, command correctness, and consistency with generated schemas.
Files:
docs/integrations/codex.mdx
**/README.md
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Update an adapter or example
README.mdwhen that adapter or example surface changes.
Files:
adapters/codex/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 Fabric contracts.
Files:
adapters/codex/README.mdadapters/codex/pyproject.tomladapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{rs,toml}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.{rs,toml}: Rust code must be formatted withcargo fmt --all; formatting can be checked withcargo fmt --all -- --check, and Rust workspaces must compile withcargo check --workspace --locked.
Rust files must begin with the specified//SPDX copyright and Apache-2.0 license header.When Rust code or Rust project configuration changes, run
cargo fmt --all -- --checkandjust test-rust.
Files:
adapters/codex/pyproject.toml
**/*.{toml,yaml,yml,sh,bash}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
TOML, YAML, and shell files must use the specified SPDX header with
#comments.
Files:
adapters/codex/pyproject.toml
**/pyproject.toml
📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)
Run
just build-pythonto verify that all Python package metadata resolves.
Files:
adapters/codex/pyproject.toml
**/*.{rs,py,pyi,toml}
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When the PyO3 bridge or package metadata changes, run
just build-pythonandcargo check -p fabric-python --locked.
Files:
adapters/codex/pyproject.tomladapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.py
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.py: Python public APIs must use type annotations, and native Python binding declarations must remain synchronized with their Rust implementations.
Python files must begin with the specified#SPDX copyright and Apache-2.0 license header.
Files:
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{rs,py}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.{rs,py}: Usesnake_casefor Rust and Python functions and variables; usePascalCasefor Rust types and Python classes.
Run tests for every language surface affected by a change. Changes touching the Rust core or public schemas require both Rust and Python test suites.
Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.
Files:
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{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/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.{py,pyi}
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When Python code or a Python-facing adapter changes, run
just test-python.
Files:
adapters/codex/src/nemo_fabric_adapters/codex/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/codex/src/nemo_fabric_adapters/codex/adapter.py
🔇 Additional comments (4)
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py (1)
281-284: LGTM!adapters/codex/pyproject.toml (1)
29-29: 🎯 Functional CorrectnessVerify the SDK bump against runtime and seam compatibility.
This changes the SDK from
0.1.0b3to0.144.4. The publishedopenai-codex0.144.4 metadata requires Python>=3.10; confirm the adapter does not still support older Python versions. (pypi.org)Also run the skill-registration integration path against the resolved artifact because the adapter relies on the SDK’s private
_client.request(...)API. As per coding guidelines, runjust build-pythonandjust test-python.Source: Coding guidelines
ATTRIBUTIONS-Python.md (1)
5890-5890: LGTM!Also applies to: 5900-5900
adapters/codex/README.md (1)
52-60: LGTM!Also applies to: 81-93, 103-104, 169-171
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
Salonijain27
left a comment
Salonijain27
left a comment
There was a problem hiding this comment.
Approved from a dependency point of view
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
|
Fern docs preview: https://nvidia-preview-pull-request-88.docs.buildwithfern.com/nemo/fabric |
Signed-off-by: Ajay Thorve <athorve@nvidia.com>
|
/merge |
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (3)
adapters/codex/README.md (1)
90-96: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winDocument the skill-directory name-uniqueness constraint.
The adapter rejects duplicate resolved skill directory names (
_native_skill_pathsinadapter.py), but this constraint isn't mentioned here. Calling it out would help users interpret thecodex_invalid_configurationerror when two configuredskills.pathsentries resolve to the same directory name.✏️ Proposed addition
- `skills.paths` names skill directories that contain `SKILL.md`. The adapter registers each directory as a process-scoped Codex skill root so Codex can - select matching skills through its normal discovery behavior. + select matching skills through its normal discovery behavior. Each resolved + directory name must be unique across `skills.paths`.🤖 Prompt for 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. In `@adapters/codex/README.md` around lines 90 - 96, Update the skills.paths documentation to state that resolved skill directory names must be unique; duplicate names are rejected by _native_skill_paths and produce a codex_invalid_configuration error. Keep the existing description of skill discovery unchanged.adapters/codex/src/nemo_fabric_adapters/codex/adapter.py (2)
29-29: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick winAvoid the private
_client.requestseam.skills/extraRoots/setstill goes through an internal SDK protocol path, so a futureopenai-codexrelease can break this without any public deprecation path. Add a narrow compatibility wrapper or test around the call.🤖 Prompt for 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. In `@adapters/codex/src/nemo_fabric_adapters/codex/adapter.py` at line 29, Update the code using SkillsExtraRootsSetResponse so the skills/extraRoots/set request no longer relies directly on the private _client.request seam. Add a narrow compatibility wrapper around this call, or add a focused test that locks the expected request behavior across openai-codex releases while keeping the existing response handling intact.
1012-1018: 🩺 Stability & Availability | 🔵 Trivial | 💤 Low valueNarrow the blind
except Exceptionin SDK cleanup.Static analysis flags catching a blind
Exceptionhere. The turn-execution branch above already narrows to(CodexError, RuntimeError, OSError); using the same tuple in thisfinallyblock would be more consistent and avoid masking unexpected error types during cleanup while still reporting the stop failure.♻️ Proposed fix
finally: if codex is not None: try: await codex.close() - except Exception: + except (CodexError, RuntimeError, OSError): output = _failure( "codex_sdk_stop_failed", "Codex SDK runtime failed to stop" )🤖 Prompt for 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. In `@adapters/codex/src/nemo_fabric_adapters/codex/adapter.py` around lines 1012 - 1018, Update the cleanup exception handler in the codex shutdown flow to catch only CodexError, RuntimeError, and OSError, matching the turn-execution handling above. Keep the existing _failure("codex_sdk_stop_failed", ...) response for those expected cleanup failures and allow unexpected exception types to propagate.Source: Linters/SAST tools
🤖 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.
Outside diff comments:
In `@adapters/codex/README.md`:
- Around line 90-96: Update the skills.paths documentation to state that
resolved skill directory names must be unique; duplicate names are rejected by
_native_skill_paths and produce a codex_invalid_configuration error. Keep the
existing description of skill discovery unchanged.
In `@adapters/codex/src/nemo_fabric_adapters/codex/adapter.py`:
- Line 29: Update the code using SkillsExtraRootsSetResponse so the
skills/extraRoots/set request no longer relies directly on the private
_client.request seam. Add a narrow compatibility wrapper around this call, or
add a focused test that locks the expected request behavior across openai-codex
releases while keeping the existing response handling intact.
- Around line 1012-1018: Update the cleanup exception handler in the codex
shutdown flow to catch only CodexError, RuntimeError, and OSError, matching the
turn-execution handling above. Keep the existing
_failure("codex_sdk_stop_failed", ...) response for those expected cleanup
failures and allow unexpected exception types to propagate.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Enterprise
Run ID: 7d7e0e3c-9d87-4ca0-920f-6152060369c7
⛔ Files ignored due to path filters (1)
uv.lockis excluded by!**/*.lock
📒 Files selected for processing (5)
adapters/codex/README.mdadapters/codex/src/nemo_fabric_adapters/codex/adapter.pydocs/integrations/codex.mdxpyproject.tomltests/adapters/test_codex_adapter.py
📜 Review details
⏰ Context from checks skipped due to timeout. (10)
- GitHub Check: Preview docs
- GitHub Check: Pre-commit
- GitHub Check: Test (Python 3.12, x86_64)
- GitHub Check: Test (Python 3.13, x86_64)
- GitHub Check: Test (Python 3.14, x86_64)
- GitHub Check: Test (Python 3.13, arm64)
- GitHub Check: Test (Python 3.12, arm64)
- GitHub Check: Test (Python 3.11, arm64)
- GitHub Check: Test (Python 3.11, x86_64)
- GitHub Check: Test (Python 3.14, arm64)
⚠️ CI failures not shown inline (6)
GitHub Actions: Rust / Test (arm64): feat: add normalized MCP and skills to Codex
Conclusion: failure
##[group]Run bail() {
�[36;1mbail() {�[0m
�[36;1m printf '::error::install-action: %s\n' "$*"�[0m
GitHub Actions: Rust / Test (x86_64): feat: add normalized MCP and skills to Codex
Conclusion: failure
##[group]Run just test-rust
�[36;1mjust test-rust�[0m
shell: /usr/bin/bash --noprofile --norc -e -o pipefail {0}
env:
CARGO_INCREMENTAL: 0
CARGO_PROFILE_DEV_DEBUG: 0
CARGO_TERM_COLOR: always
RUST_BACKTRACE: short
RUSTFLAGS: -D warnings
CARGO_UNSTABLE_SPARSE_REGISTRY: true
CARGO_REGISTRIES_CRATES_IO_PROTOCOL: sparse
CACHE_ON_FAILURE: false
##[endgroup]
cargo test --workspace --locked
�[1m�[92m Compiling�[0m nemo-fabric-core v0.1.0 (/home/runner/work/NeMo-Fabric/NeMo-Fabric/crates/fabric-core)
�[1m�[92m Compiling�[0m nemo-fabric-cli v0.1.0 (/home/runner/work/NeMo-Fabric/NeMo-Fabric/crates/fabric-cli)
�[1m�[92m Finished�[0m `test` profile [unoptimized] target(s) in 4.99s
�[1m�[92m Running�[0m unittests src/lib.rs (target/debug/deps/nemo_fabric_cli-c6ca16d34f22a9db)
running 19 tests
Experiment with NeMo Fabric presets and maintained examples
Usage: nemo-fabric <COMMAND>
Commands:
preset Discover complete built-in presets
example Discover and initialize maintained examples
plan Resolve and print a run plan as JSON
doctor Diagnose a selected configuration without running it
run Execute one request through a selected configuration
version Print the Fabric core version
help Print this message or the help of the given subcommand(s)
Options:
-h, --help Print help
-V, --version Print version
test app::tests::explicit_help_is_successful ... ok
test app::tests::failed_and_cancelled_runs_are_errors ... ok
test app::tests::missing_subcommand_is_an_error ... ok
test app::tests::model_overrides_require_a_preset_with_a_default_model ... ok
test app::tests::parses_example_variant ... ok
test app::tests::parses_preset_plan ... ok
test app::tests::plan_requires_a_preset ... ok
test app::tests::preset_model_overrides_preserve_provider_settings ... ok
test assets::tests::stages_and_removes_embedded_files ... ok
test app::tests::selectors_are_mutually_exclusive ... ok
test exam...
GitHub Actions: Rust / Test (arm64): feat: add normalized MCP and skills to Codex
Conclusion: failure
##[group]Run just test-rust
�[36;1mjust test-rust�[0m
shell: /usr/bin/bash --noprofile --norc -e -o pipefail {0}
env:
CARGO_INCREMENTAL: 0
CARGO_PROFILE_DEV_DEBUG: 0
CARGO_TERM_COLOR: always
RUST_BACKTRACE: short
RUSTFLAGS: -D warnings
CARGO_UNSTABLE_SPARSE_REGISTRY: true
CARGO_REGISTRIES_CRATES_IO_PROTOCOL: sparse
CACHE_ON_FAILURE: false
##[endgroup]
cargo test --workspace --locked
�[1m�[92m Compiling�[0m nemo-fabric-core v0.1.0 (/home/runner/work/NeMo-Fabric/NeMo-Fabric/crates/fabric-core)
�[1m�[92m Compiling�[0m nemo-fabric-cli v0.1.0 (/home/runner/work/NeMo-Fabric/NeMo-Fabric/crates/fabric-cli)
�[1m�[92m Finished�[0m `test` profile [unoptimized] target(s) in 5.49s
�[1m�[92m Running�[0m unittests src/lib.rs (target/debug/deps/nemo_fabric_cli-14421df198e0c90a)
running 19 tests
Experiment with NeMo Fabric presets and maintained examples
Usage: nemo-fabric <COMMAND>
Commands:
preset Discover complete built-in presets
example Discover and initialize maintained examples
plan Resolve and print a run plan as JSON
doctor Diagnose a selected configuration without running it
run Execute one request through a selected configuration
version Print the Fabric core version
help Print this message or the help of the given subcommand(s)
Options:
-h, --help Print help
-V, --version Print version
test app::tests::explicit_help_is_successful ... ok
test app::tests::failed_and_cancelled_runs_are_errors ... ok
test app::tests::missing_subcommand_is_an_error ... ok
test app::tests::model_overrides_require_a_preset_with_a_default_model ... ok
test app::tests::parses_example_variant ... ok
test app::tests::parses_preset_plan ... ok
test app::tests::plan_requires_a_preset ... ok
test app::tests::preset_model_overrides_preserve_provider_settings ... ok
test app::tests::selectors_are_mutually_exclusive ... ok
test assets::tests::stages_and_removes_embedded_files ... ok
test exam...
GitHub Actions: Rust / 1_Test (arm64).txt: feat: add normalized MCP and skills to Codex
Conclusion: failure
##[group]Run bail() {
�[36;1mbail() {�[0m
�[36;1m printf '::error::install-action: %s\n' "$*"�[0m
GitHub Actions: Rust / Test (x86_64): feat: add normalized MCP and skills to Codex
Conclusion: failure
##[group]Run bail() {
�[36;1mbail() {�[0m
�[36;1m printf '::error::install-action: %s\n' "$*"�[0m
GitHub Actions: Rust / 0_Test (x86_64).txt: feat: add normalized MCP and skills to Codex
Conclusion: failure
##[group]Run bail() {
�[36;1mbail() {�[0m
�[36;1m printf '::error::install-action: %s\n' "$*"�[0m
🧰 Additional context used
📓 Path-based instructions (33)
**/*.{rs,toml}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.{rs,toml}: Rust code must be formatted withcargo fmt --all; formatting can be checked withcargo fmt --all -- --check, and Rust workspaces must compile withcargo check --workspace --locked.
Rust files must begin with the specified//SPDX copyright and Apache-2.0 license header.When Rust code or Rust project configuration changes, run
cargo fmt --all -- --checkandjust test-rust.
Files:
pyproject.toml
**/*
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*: All source files must include the specified SPDX copyright and Apache-2.0 license header using the comment syntax appropriate to the file type.
Release tags must use raw Rust-compatible SemVer without a leadingv, such as0.1.0or0.1.0-rc.1.
**/*: 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.
**/*: Keep pull request branch scope coherent and reviewable.
Run relevant tests undervalidate-changebefore opening or updating a pull request.
Format changed files with the language-native formatter.
Update documentation and examples for public behavior changes.
Update dependent maintainer or consumer guidance when code changes affect APIs, bindings, commands, paths, packaging guidance, or best practices.
Use Conventional Commit style for pull request titles:<type>: <concise imperative summary>, choosing the type from the actual change surface. Usefixonly for user-facing or runtime product-code bug fixes.
A pull request body must include#### Overview,#### Details,#### Validation,#### Where should the reviewer start?, and `#### Related ...
Files:
pyproject.tomldocs/integrations/codex.mdxadapters/codex/README.mdadapters/codex/src/nemo_fabric_adapters/codex/adapter.pytests/adapters/test_codex_adapter.py
**/*.{toml,yaml,yml,sh,bash}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
TOML, YAML, and shell files must use the specified SPDX header with
#comments.
Files:
pyproject.toml
{Cargo.toml,pyproject.toml,python/pyproject.toml,python/src/nemo_fabric/**}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
{Cargo.toml,pyproject.toml,python/pyproject.toml,python/src/nemo_fabric/**}: Keep Rust package names, Python package/import paths, and native module names internally consistent across Cargo and Python packaging metadata and source paths.
Ensure generated native and Python artifacts are placed where downstream consumers expect them.
Files:
pyproject.toml
{Cargo.toml,pyproject.toml,python/pyproject.toml,Cargo.lock,uv.lock}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Keep workspace, Rust, Python, and lockfile versions aligned where required.
Files:
pyproject.toml
{pyproject.toml,python/pyproject.toml,Cargo.toml,python/src/nemo_fabric/**}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
The editable maturin build must continue to produce the
nemo_fabric._nativeextension.
Files:
pyproject.toml
{.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml,justfile,pyproject.toml,python/pyproject.toml,Cargo.toml}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
CI workflows must reference the same package names and commands used by local build and installation workflows.
Files:
pyproject.toml
{Cargo.toml,Cargo.lock,pyproject.toml,python/pyproject.toml,uv.lock,docs/package.json,docs/package-lock.json}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Maintain package metadata and dependency resolution consistently across Rust, Python, documentation tooling, and their lockfiles.
Files:
pyproject.toml
**/pyproject.toml
📄 CodeRabbit inference engine (.agents/skills/update-project-version/SKILL.md)
Run
just build-pythonto verify that all Python package metadata resolves.
Files:
pyproject.toml
**/*.{rs,py,pyi,toml}
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When the PyO3 bridge or package metadata changes, run
just build-pythonandcargo check -p fabric-python --locked.
Files:
pyproject.tomladapters/codex/src/nemo_fabric_adapters/codex/adapter.pytests/adapters/test_codex_adapter.py
**/*.{md,mdx,html}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Changes affecting public behavior, adapters, examples, or workspace structure must update the corresponding documentation; public API changes require updated SDK or API reference documentation.
Files:
docs/integrations/codex.mdxadapters/codex/README.md
**/*.{md,mdx}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
For docs site changes, run
just docsto regenerate Python and Rust API references and validate Fern configuration.
**/*.{md,mdx}: Prioritize factual accuracy in NeMo Fabric documentation and keep commands, package names, APIs, file paths, repository layout, entry points, support claims, examples, and procedures aligned with current repository behavior.
Update relevant entry-point documentation when public behavior changes, includingREADME.md,docs/index.yml, package or crate READMEs, and adapter or integration READMEs.
Use{/* ... */}delimiters for top-of-file SPDX comments in MDX files, not HTML comment delimiters.
CapitalizeNVIDIAcorrectly and use consistent current repository terminology, product names, APIs, and feature names.
Format commands, code, expressions, file names, paths, and filenames as inline code where appropriate.
Use title case for technical-documentation headings.
Introduce code blocks, tables, and lists with complete lead-in sentences.
Use descriptive link text instead of raw URLs or generic labels such ashere.
Write procedures as short, imperative, parallel, easy-to-scan steps; prefer active voice, present tense, plain English, and concise sentences.
Useafterinstead ofoncewhen expressing temporal sequence, and usecaninstead ofmaywhen describing possibility rather than permission.
Use unambiguous date formats and avoid ordinal dates in body text.
When reviewing documentation, report findings in severity order underMust fix,Should fix, andNice to have, with file paths, line references, explanations, and concrete rewrites or directions.
Files:
docs/integrations/codex.mdxadapters/codex/README.md
**/*.mdx
📄 CodeRabbit inference engine (CONTRIBUTING.md)
MDX files must use the specified SPDX header in a JSX comment.
In MDX files, use JSX comment delimiters (
{/*and*/}) for top-of-file comments, including SPDX headers; do not use HTML comments.
Files:
docs/integrations/codex.mdx
{README.md,docs/**/*.{md,mdx,yml},examples/**/*.{md,mdx,yml}}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Keep package names, repository references, and build commands current in documentation and examples.
Files:
docs/integrations/codex.mdx
{docs/**/*.{md,mdx,yml},examples/**/*.{md,mdx,yml}}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Update relevant getting-started, reference, adapter, and example documentation when the corresponding examples or adapters change.
Files:
docs/integrations/codex.mdx
docs/**/*.{md,mdx,yml}
📄 CodeRabbit inference engine (.agents/skills/contribute-docs/SKILL.md)
Run
just docswhen the documentation site changes.
Files:
docs/integrations/codex.mdx
{docs/**/*,.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml,justfile}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Use the current install, import, build, test, clean, and documentation commands consistently in documentation, examples, CI workflows, and just recipes.
Files:
docs/integrations/codex.mdx
{docs/**/*,.github/workflows/ci_python.yml,.github/workflows/ci_rust.yml}
📄 CodeRabbit inference engine (.agents/skills/maintain-packaging/SKILL.md)
Reflect public packaging changes in release-facing documentation and examples.
Files:
docs/integrations/codex.mdx
**/*.{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 spellNVIDIAin all caps; do not useNvidia,nvidia, orNV.
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 ashereorread 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.
Usecanfor possibility and reservemayfor permission.
Useafterfor temporal relationships instead ofonce, and preferrefer tooverseewhen 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:
docs/integrations/codex.mdxadapters/codex/README.md
docs/**/*
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
For documentation or examples changes, run
just docswhen practical and verify documented commands against the current repository.
Files:
docs/integrations/codex.mdx
{docs/**,README.md,AGENTS.md}
⚙️ CodeRabbit configuration file
{docs/**,README.md,AGENTS.md}: Review documentation for technical accuracy against the current API, command correctness, and consistency with generated schemas.
Files:
docs/integrations/codex.mdx
**/README.md
📄 CodeRabbit inference engine (CONTRIBUTING.md)
Update an adapter or example
README.mdwhen that adapter or example surface changes.
Files:
adapters/codex/README.md
**/*.{html,md}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
HTML and Markdown files must use the specified SPDX header in an HTML comment.
Files:
adapters/codex/README.md
**/*.{md,rst}
📄 CodeRabbit inference engine (.agents/skills/contribute-api/SKILL.md)
Update documentation and examples in the same branch as the public API change.
Verify README and documentation entry points, package names, paths, examples, and public commands remain current after changes.
Files:
adapters/codex/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.
Usecanfor possibility and reservemayfor permission; useafterfor temporal order; userefer tofor cross-references; prefer short direct sentences and specific verbs; avoid unnecessarypleasein 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: usefor exampleorsuch asinstead ofe.g.,and so oninstead ofetc.,that isinstead ofi.e.,compared toinstead ofvs., andby,through, orusinginstead ofvia. Use industry-standard terms such as in silico, in vitro, and in vivo when appropriate, and italicize them in running text.
Usethatwithout commas for essential clauses, andwhichwith commas for nonessential clauses.
Format dates and times clearly: spell out months in body text; use forms such asJune 12, 2025; avoid numeric or ordinal dates; capitalize days; use 12-hour time when appropriate; include a space beforea.m.orp.m.; useETandPTfor needed time zones; avoid24/7; and preferfrom 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/codex/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 Fabric contracts.
Files:
adapters/codex/README.mdadapters/codex/src/nemo_fabric_adapters/codex/adapter.py
**/*.py
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.py: Python public APIs must use type annotations, and native Python binding declarations must remain synchronized with their Rust implementations.
Python files must begin with the specified#SPDX copyright and Apache-2.0 license header.
Files:
adapters/codex/src/nemo_fabric_adapters/codex/adapter.pytests/adapters/test_codex_adapter.py
**/*.{rs,py}
📄 CodeRabbit inference engine (CONTRIBUTING.md)
**/*.{rs,py}: Usesnake_casefor Rust and Python functions and variables; usePascalCasefor Rust types and Python classes.
Run tests for every language surface affected by a change. Changes touching the Rust core or public schemas require both Rust and Python test suites.
Public contract changes must keep native Python binding declarations synchronized with their Rust implementations.
Files:
adapters/codex/src/nemo_fabric_adapters/codex/adapter.pytests/adapters/test_codex_adapter.py
**/*.{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/codex/src/nemo_fabric_adapters/codex/adapter.pytests/adapters/test_codex_adapter.py
**/*.{py,pyi}
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When Python code or a Python-facing adapter changes, run
just test-python.
Files:
adapters/codex/src/nemo_fabric_adapters/codex/adapter.pytests/adapters/test_codex_adapter.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.asyncioto tests; async tests are automatically detected by the async runner.
Do not add-> Nonereturn annotations to test functions.
When mocking a class, useunittest.mock.MagicMockorAsyncMock, supplyingspecwhen necessary; do not define a new mock class.
Prefix mocked class names withmock, notfake.
Prefer pytest fixtures over helper methods.
Define shared fixtures inconftest.pyrather than repeating them across test files.
Define fixtures using@pytest.fixture(name="<fixture_name>"[, scope="<scope>"])and a<fixture_name>_fixturefunction; specifyscopeonly when it is notfunction.
Preferpytest.mark.parametrizeover separate tests for different input types.
Use@pytest.mark.usefixtureswhen a fixture is needed but its return value is unused.
Useos.environto modify environment variables in tests; do not usemonkeypatch.setenv, because the autouserestore_environ_fixtureintests/conftest.pyrestores the environment after each test.
Avoid defensive programming in tests; access expected data directly so missing data raises a clear error instead of being silently tolerated.
Run focused tests withuv run pytest -k "<pattern>"and all tests withuv run pytest.
Files:
tests/adapters/test_codex_adapter.py
tests/adapters/**/*
📄 CodeRabbit inference engine (.agents/skills/validate-change/SKILL.md)
When an adapter or integration changes, run its focused tests under
tests/adapters, followed byjust test-python.
Files:
tests/adapters/test_codex_adapter.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_codex_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/codex/src/nemo_fabric_adapters/codex/adapter.py
🧬 Code graph analysis (2)
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py (2)
adapters/common/src/nemo_fabric_adapters/common/utils.py (2)
base_dir(62-68)relay_enabled(156-157)tests/adapters/test_codex_adapter.py (1)
close(119-120)
tests/adapters/test_codex_adapter.py (2)
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py (2)
run(1098-1108)state_dir(490-491)adapters/common/src/nemo_fabric_adapters/common/utils.py (1)
base_dir(62-68)
🪛 Ruff (0.15.21)
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py
[warning] 1015-1015: Do not catch blind exception: Exception
(BLE001)
tests/adapters/test_codex_adapter.py
[warning] 488-488: Missing return type annotation for private function fail_to_create_home
Add return type annotation: Never
(ANN202)
[warning] 488-488: Missing type annotation for *_args
(ANN002)
[warning] 488-488: Missing type annotation for **_kwargs
(ANN003)
[warning] 489-489: Avoid specifying long messages outside the exception class
(TRY003)
🔇 Additional comments (8)
adapters/codex/src/nemo_fabric_adapters/codex/adapter.py (3)
241-274: Falsyskill_pathsfix confirmed.
.get("skill_paths", [])now defaults only when the key is absent, preserving the intended validation for falsy non-list values. This matches the previously requested fix.
253-253: Consistentbase_dir-relative path resolution.All native path resolution (
skill_paths,resolve_cwd,codex_bin, artifact root, relay command base) consistently resolves againstcommon_utils.base_dir(payload)withoutexpanduser()or reliance onharness.settings.cwd.
Based on learnings, this matches the required convention for adapter-configured paths.Also applies to: 300-308, 428-478, 481-486, 634-647, 731-747
689-698: MCP/skills wiring and validation ordering look correct.
thread_configinjects nativemcp_serversonly when present, config overrides are merged afterward,validate_payloadresolvesskill_pathsup-front and exercisesthread_config(which validates MCP servers) before touching the SDK, andinvoke_codex_sdkregisters skill roots before thread creation inside the timeout boundary — matching the PR's stated ordering guarantees.Also applies to: 781-813, 943-969
pyproject.toml (1)
128-128: CLI-bin constraint removal matches the SDK dependency retention plan.Removing the
openai-codex-cli-bin==0.137.0a4constraint aligns with retaining the SDK's own exact transitive pin, andopenai-codex==0.144.4is a valid, current release line (confirmed against the upstreamopenai/codexrelease history).adapters/codex/README.md (1)
52-69: Documentation matches current adapter behavior.The NVIDIA provider,
codex_binresolution, MCP/skills mapping,tools.blockedunsupported rationale, Relay-provider restriction, and Relay observability guidance all match the corresponding code inadapter.pyand are consistent withdocs/integrations/codex.mdx.Also applies to: 98-103, 112-113, 150-156, 169-172
docs/integrations/codex.mdx (1)
104-132: Relay/NVIDIA provider restriction is consistent with README and adapter code.The Relay-requires-
openai-provider statement matchesvalidate_payload's check and the equivalent README section.tests/adapters/test_codex_adapter.py (2)
109-116: Mock setup and new MCP-mapping/no-call assertions look correct.The
_client.requestmock seam correctly exercises the skill-registration compatibility path, and the oneshot test'sclient._client.request.assert_not_awaited()/client.models.assert_not_awaited()assertions properly guard against unnecessary calls when no MCP/skills config is present.os.environusage andAsyncMockfollow the repository's test conventions.Also applies to: 194-231
819-828: Regression coverage for falsy capability settings and resolved plan shape is solid.Rejecting
{}formcp_servers/skillsand asserting the resolved native capability plan (skill_paths,mcp_serverswithexposure) gives good end-to-end coverage of the normalized configuration contract.Also applies to: 862-889
| description: "Configure ChatGPT or API key authentication for the NeMo Fabric Codex SDK adapter." | ||
| title: "Use the Codex Adapter" | ||
| description: "Configure capabilities, authentication, and telemetry for the NeMo Fabric Codex SDK adapter." | ||
| --- |
There was a problem hiding this comment.
Must fix: This is the first product-name use on the page, but it omits the NVIDIA prefix. The documentation style guide requires “NVIDIA NeMo Fabric” on first use, followed by “NeMo Fabric.” Please change the description to “Configure capabilities, authentication, and telemetry for the NVIDIA NeMo Fabric Codex SDK adapter.”
4 participants
Overview
Close the remaining truthful Codex capability gaps against the Claude adapter. This PR adds normalized MCP and skill support without claiming partial
tools.blockedenforcement. It also completes FABRIC-100 by moving the adapter to the stable Codex SDK/runtime versioning contract: Fabric directly pinsopenai-codex==0.144.4, while the SDK owns its exactopenai-codex-cli-bin==0.144.4transitive dependency.Fabric no longer declares or constrains the app-server package separately. The built Codex adapter wheel contains no direct CLI-bin requirement. No dependency is added or removed; both Codex packages remain Apache-2.0. This PR does not change the public Fabric API or schema.
Details
FabricConfig.mcpentries into request-scoped Codexmcp_serversconfiguration for stdio, HTTP, and streamable HTTP transports. The stdio mapping uses Python's standard-libraryshlexmodule.FabricConfig.skills.pathswith the Codex app-server before thread creation while preserving Codex-native implicit skill selection.skill_pathsvalues before SDK startup and preserve SDK cleanup when the private skill-registration seam is unavailable.mcpandskillsin the adapter descriptor and document their public Python SDK usage.tools.blockedexplicitly unsupported. The pinned Codex runtime does not expose one deny boundary across built-in, local, MCP, and hosted tools.openai-codex==0.1.0b3toopenai-codex==0.144.4, remove Fabric-owned CLI-bin constraints, regenerate both lockfiles, and update attribution evidence.openai-codex==0.144.4still does not wrap the app-server's process-scopedskills/extraRoots/setRPC. The adapter isolates that pinned-SDK compatibility seam in_register_skill_roots()so consumers do not need to stage skills into their workspace.Validation
just test-python— 389 passed, 11 skippeduv run --no-sync pytest -q tests/adapters/test_codex_adapter.py— 42 passed, including thread start/resume, explicit and relativecodex_binoverrides, structured-output forwarding, failed/incomplete result normalization, timeout interruption, resume identity validation, and SDK close behaviorRUN_FABRIC_CODEX_INTEGRATION=1E2E withFABRIC_TEST_CODEX_BINset to the SDK-bundled 0.144.4 executable — 1 passed in 87.42s; covered ChatGPT authentication, one-shot execution, thread start/resume with state continuity, explicit runtime override, and context-managed shutdownFabric().run()with ChatGPT authentication returned the expected{"count": 2, "status": "ok"}payload through the 0.144.4 app-serverFabric().run()with ChatGPT authentication, an implicitly selected temporary skill, and the remote OpenAI Docs MCP server returned the expected skill marker and MCP-derived answerRUN_FABRIC_CODEX_RELAY_INTEGRATION=1 ... pytest -q tests/e2e/test_codex.py::test_codex_sdk_with_relay -swithnemo-relay 0.6.0— 1 passed; verified gateway routing, ATIF/ATOF telemetry, thread resume, and state continuityjust --set no_uv true build-python— passeduv build --project adapters/codex --out-dir /tmp/nemo-fabric-pr88-codex-wheel— passed; wheel metadata requiresopenai-codex==0.144.4and has no directopenai-codex-cli-binrequirementcargo check -p fabric-python --locked— passeduv lock --check --project .anduv lock --check --project adapters/codex— passeduv run --no-project python scripts/licensing/license_diff.py --base-ref origin/main— no packages added or removed; both Codex packages remain Apache-2.0just docswith the bundled Node runtime — 0 Fern errors, 1 existing warningWhere should the reviewer start?
Start with
_native_mcp_servers(),_native_skill_paths(), and_register_skill_roots()inadapters/codex/src/nemo_fabric_adapters/codex/adapter.py. Then review the behavior-focused coverage intests/adapters/test_codex_adapter.py, the exact advertised capability list inadapters/codex/fabric-adapter.json, and the dependency ownership change inadapters/codex/pyproject.toml.The key design decisions are intentionally narrow: this PR does not claim
tools.blockedparity until Codex can enforce the normalized deny policy across every tool surface, and Fabric depends only on the SDK rather than separately owning the SDK's app-server distribution.Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)
Closes FABRIC-100
Relates to FABRIC-50
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
mcp/skillsnow fail fast with clearcodex_invalid_configurationerrors.codex_binoverride, and notedtools.blockedis unsupported.