Skip to content

Add cuda-python agent skill for users and evals - #3452

Open
danielfrg wants to merge 5 commits into
NVIDIA:mainfrom
danielfrg:ai-skills
Open

danielfrg wants to merge 5 commits into
NVIDIA:mainfrom
danielfrg:ai-skills

Conversation

@danielfrg

Copy link
Copy Markdown
Contributor

Add skills/cuda-python/SKILL.md covering cuda.core, cuda.bindings and cuda.pathfinder, with 20 eval cases and their starter files.

Description

closes

Checklist

  • I have read the Contributing Guidelines.
  • New or existing tests cover these changes.
  • The documentation is up to date with these changes.

Add skills/cuda-python/SKILL.md covering cuda.core, cuda.bindings and
cuda.pathfinder, with 20 eval cases and their starter files.
@copy-pr-bot

copy-pr-bot Bot commented Nov 13, 2026

Copy link
Copy Markdown
Contributor

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@danielfrg danielfrg self-assigned this Nov 13, 2026
@danielfrg danielfrg added the P1 Medium priority - Should do label Nov 13, 2026
@danielfrg danielfrg added this to the cuda.core next milestone Nov 13, 2026
@danielfrg danielfrg added documentation Improvements or additions to documentation cuda.bindings Everything related to the cuda.bindings module cuda.core Everything related to the cuda.core module cuda.pathfinder Everything related to the cuda.pathfinder module labels Nov 13, 2026
@coderabbitai

coderabbitai Bot commented Nov 13, 2026 •

Copy link
Copy Markdown

Explain Change →

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

No actionable comments were generated in the recent review.

Summary

Summary by CodeRabbit

  • Documentation
    • Added a CUDA Python guide with installation guidance, API recommendations, and examples for compiling and launching kernels.
    • The guide covers kernel arguments, header discovery, interoperability with CuPy, PyTorch, and Numba, CUDA graph capture, and migration from older CUDA Python and PyCUDA APIs.
    • Added examples covering performance measurement, data conversion, error handling, scalar arguments, stream coordination, and legacy API workflows.
    • Added skill metadata and a benchmark report describing supported task areas, evaluation results, and scoring methodology.
  • Tests
    • Added evaluation cases covering CUDA Python workflows, migrations, and task classification.

The change adds a CUDA Python guide, skill metadata, benchmark documentation, 16 evaluation cases, and runnable examples. The examples cover CUDA Core, legacy bindings, and PyCUDA workflows.

Changes: CUDA Python skill and evaluations

Layer / File(s) Summary
Guide and skill scope
skills/cuda-python/SKILL.md, skills/cuda-python/skill-card.md, skills/cuda-python/BENCHMARK.md, skills/cuda-python/evals/evals.json, skills/cuda-python/skill.oms.sig, toolshed/check_spdx.py, .pre-commit-config.yaml
The guide covers installation, cuda.core workflows, framework interoperability, graph capture, header discovery, and migration. The skill card and benchmark describe the skill scope and evaluation results. The evaluation suite defines 16 cases. The SPDX map assigns Apache-2.0 to skills; the signature bundle attests to the skill digest and related file digests. The pre-commit exclusions now match skill signature files.
Core and runtime examples
skills/cuda-python/evals/files/cuda-python-fix-hardcoded-paths/code/half_roundtrip.py, skills/cuda-python/evals/files/cuda-python-fix-ignored-errors/code/hot_cells.py, skills/cuda-python/evals/files/cuda-python-fix-scalar-args/code/normalize.py, skills/cuda-python/evals/files/cuda-python-fix-stream-order/code/denoise.py, skills/cuda-python/evals/files/cuda-python-perf-launch-loop/code/diffuse.py, skills/cuda-python/evals/files/cuda-python-review-calibrate/code/calibrate.py
The examples compile and launch CUDA kernels for numerical processing, counting, normalization, stream interoperability, diffusion, and calibration. They synchronize work and report results as described in each example.
API migration examples
skills/cuda-python/evals/files/cuda-python-port-core-experimental/code/pipeline.py, skills/cuda-python/evals/files/cuda-python-port-legacy-bindings/code/saxpy_legacy.py, skills/cuda-python/evals/files/cuda-python-port-pycuda/code/threshold_pycuda.py
The examples demonstrate CUDA Core experimental LTO linking, CUDA 11.8 low-level bindings, and PyCUDA kernel workflows.

Suggested reviewers: seberg, juenglin

Change: Other

Merge Risk: Moderate · up to c6950

The new CUDA Python skill guide and eval examples still contain several snippets that can fail or behave unsafely on some hardware or library versions. Correct them before merge so the guidance users and evals rely on is accurate. The signature file no longer blocks the pre-commit hooks.

Architecture risk: Low · up to c6950

The change affects 2 systems.

Changed systems: skills, toolshed

Architecture concerns
No architecture-level concerns identified.

Review details

Systems and components

  • observed — skills (service) was modified; 14 changed files map to changed impact.
  • observed — toolshed (service) was modified; 1 changed file maps to changed impact.

Before / after behavior

  • observed — Modified behavior in skills/cuda-python/SKILL.md: Adds a CUDA Python guide covering the skill’s scope and exclusions, package installation, cuda.core API examples, kernel compilation and launch, argument types, third-party stream interoperability, CUDA graph capture, cuda.pathfinder, migration mappings, and references. The guide notes that stream work is asynchronous until synchronization, launch() does not check kernel signatures, and a separate CUDA stream is not ordered with another library’s stream.
  • observed — Modified behavior in skills/cuda-python/evals/evals.json: Adds a JSON evaluation suite containing 16 CUDA Python prompts and expected-output criteria, with assertions for each. Fifteen cases target cuda-python; the CuPy-only and PyTorch-install negative cases set expected_skill to null. The suite covers implementation, migration, debugging, performance, API usage, and installation; it does not add runtime code.
  • observed — Modified behavior in skills/cuda-python/evals/files/cuda-python-fix-hardcoded-paths/code/half_roundtrip.py: Defines a fixed CUDA include path and a kernel that selects one of four scale factors, applies it to each in-range weight, and stores the absolute difference between the scaled float32 value and its half-precision round trip.
  • observed — Modified behavior in skills/cuda-python/evals/files/cuda-python-fix-hardcoded-paths/code/half_roundtrip.py: Adds main() to create a CUDA device and stream, compile the kernel, and process 1,048,576 float32 values from -1 to 1. For each of four layers, it launches the kernel, synchronizes, and prints the maximum error to six decimal places; the script invokes main() when run directly.
- [ ] Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Recent review info

Run configuration

Commits

Reviewing files that changed from the base of the PR and between 01d7742 and c695018.

Files selected for processing: 1

  • .pre-commit-config.yaml

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 11 remain after this review.

@coderabbitai coderabbitai Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actionable comments posted: 5

Additional comments: 1

skills/cuda-python/SKILL.md-86-86: 1

86-86: Major · Stability & Availability · Quick win

Unverified finding
Verification ran but could not confirm this finding. It is shown for review, not as a verified issue.

important: Wait for the host allocation before writing.

PinnedMemoryResource.allocate is stream-ordered, but this CPU write does not wait for the allocation to complete. This can access the allocation too early. Synchronize before host access, or use a synchronous pinned allocation. (raw.githubusercontent.com)

Review info

Run configuration

Commits

Reviewing files that changed from the base of the PR and between b715078 and 6f44676.

Files selected for processing: 11

  • skills/cuda-python/SKILL.md
  • skills/cuda-python/evals/evals.json
  • skills/cuda-python/evals/files/cuda-python-fix-hardcoded-paths/code/half_roundtrip.py
  • skills/cuda-python/evals/files/cuda-python-fix-ignored-errors/code/hot_cells.py
  • skills/cuda-python/evals/files/cuda-python-fix-scalar-args/code/normalize.py
  • skills/cuda-python/evals/files/cuda-python-fix-stream-order/code/denoise.py
  • skills/cuda-python/evals/files/cuda-python-perf-launch-loop/code/diffuse.py
  • skills/cuda-python/evals/files/cuda-python-port-core-experimental/code/pipeline.py
  • skills/cuda-python/evals/files/cuda-python-port-legacy-bindings/code/saxpy_legacy.py
  • skills/cuda-python/evals/files/cuda-python-port-pycuda/code/threshold_pycuda.py
  • skills/cuda-python/evals/files/cuda-python-review-calibrate/code/calibrate.py

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 11 remain after this review.

#include <cuda_fp16.h>
#include <cuda/std/array>

constexpr cuda::std::array<float, 4> layer_scale = {1.0f, 8.0f, 64.0f, 512.0f};

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Major · Functional Correctness · Quick win

important: Make layer_scale accessible to the kernel.

CUDA 12.4 treats this namespace-scope constexpr array as a host object. half_error indexes it with the runtime value layer, so compilation fails even after the header paths are fixed. Move the array into the kernel or use device constant storage. Keep the <cuda/std/array> dependency if the evaluation must test CCCL header lookup. (docs.nvidia.com)

Comment thread skills/cuda-python/SKILL.md Resolved

# 3. Allocate pinned host memory and device memory.
n = 1 << 20
h_x = PinnedMemoryResource().allocate(n * 4, stream=stream)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Major · Functional Correctness · Quick win

important: Support devices without host memory pools.

After fixing the import, this allocation can fail on devices without host memory pool support. Use LegacyPinnedMemoryResource for this example, or check the capability and provide a fallback. (raw.githubusercontent.com)

```python
view = StridedMemoryView.from_any_interface(arr, stream_ptr=stream.handle)
c_strides = tuple(int(np.prod(view.shape[i + 1 :])) for i in range(len(view.shape)))
ok = view.is_device_accessible and view.dtype == np.float32 and view.strides in (None, c_strides)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Major · Functional Correctness · Quick win

important: Validate the array’s device.

On a multi-GPU system, an array on GPU 1 can pass this check while the kernel runs on GPU 0. is_device_accessible does not establish that GPU 0 can access the allocation. Check view.device_id against the launch device, or verify peer access before launch. (nvidia.github.io)

Comment on lines +139 to +140
include = [find_nvidia_header_directory("cudart"), find_nvidia_header_directory("cccl")] # None if missing
options = ProgramOptions(std="c++17", include_path=include)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Major · Functional Correctness · Quick win

important: Handle missing header directories.

If either lookup returns None, ProgramOptions converts it to --include-path=None. Compilation then cannot find the missing headers. Check each result and tell users which header package to install before constructing the options. (nvidia.github.io)

Write eval starter script output with sys.stdout.write, use snake_case
names in the legacy bindings starter, and register the skills directory
in the SPDX check. Use PinnedMemoryResource in the skill example.

@coderabbitai coderabbitai Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actionable comments posted: 1

Caution

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

⚠️ Outside diff range comments (1)

Major · important: Require a NumPy version that provides a writable DLPack view. · SKILL.md:85

skills/cuda-python/SKILL.md:85
Major · Functional Correctness · Quick win

important: Require a NumPy version that provides a writable DLPack view.

With NumPy 2.0, np.from_dlpack() returns a read-only view, so the assignment at Line 86 raises ValueError. The installation instructions do not require a newer NumPy version. Require a compatible version or use a writable host-memory representation. (numpy.org)

Review info

Run configuration

Commits

Reviewing files that changed from the base of the PR and between 6f44676 and 3f28bf3.

Files selected for processing: 12

  • skills/cuda-python/SKILL.md
  • skills/cuda-python/evals/evals.json
  • skills/cuda-python/evals/files/cuda-python-fix-hardcoded-paths/code/half_roundtrip.py
  • skills/cuda-python/evals/files/cuda-python-fix-ignored-errors/code/hot_cells.py
  • skills/cuda-python/evals/files/cuda-python-fix-scalar-args/code/normalize.py
  • skills/cuda-python/evals/files/cuda-python-fix-stream-order/code/denoise.py
  • skills/cuda-python/evals/files/cuda-python-perf-launch-loop/code/diffuse.py
  • skills/cuda-python/evals/files/cuda-python-port-core-experimental/code/pipeline.py
  • skills/cuda-python/evals/files/cuda-python-port-legacy-bindings/code/saxpy_legacy.py
  • skills/cuda-python/evals/files/cuda-python-port-pycuda/code/threshold_pycuda.py
  • skills/cuda-python/evals/files/cuda-python-review-calibrate/code/calibrate.py
  • toolshed/check_spdx.py
Files skipped from review as they are similar to previous changes: 8

  • skills/cuda-python/evals/files/cuda-python-fix-hardcoded-paths/code/half_roundtrip.py
  • skills/cuda-python/evals/files/cuda-python-fix-ignored-errors/code/hot_cells.py
  • skills/cuda-python/evals/files/cuda-python-fix-stream-order/code/denoise.py
  • skills/cuda-python/evals/files/cuda-python-perf-launch-loop/code/diffuse.py
  • skills/cuda-python/evals/files/cuda-python-port-core-experimental/code/pipeline.py
  • skills/cuda-python/evals/files/cuda-python-port-legacy-bindings/code/saxpy_legacy.py
  • skills/cuda-python/evals/files/cuda-python-port-pycuda/code/threshold_pycuda.py
  • skills/cuda-python/evals/files/cuda-python-review-calibrate/code/calibrate.py

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 10 remain after this review.

```python
import numpy as np

from cuda.core import Device, LaunchConfig, PinnedMemoryResource, Program, launch

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Major · Stability & Availability · Quick win

important: Synchronize the host allocation before writing to it.

PinnedMemoryResource.allocate() is asynchronous. The host write at Line 86 can access the allocation before it completes, which has undefined behavior. Synchronize the stream before accessing h_x, or use LegacyPinnedMemoryResource. (raw.githubusercontent.com)

@danielfrg danielfrg mentioned this pull request Nov 13, 2026
3 tasks done

@mdboom mdboom left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we have evidence that doing this is better than improving our docs (which would also benefit non-agent-using humans)?

@danielfrg

Copy link
Copy Markdown
Contributor Author

We should continue making our docs better for humans and agents. The skill mentions to see the docs in case of doubts, so thats ultimately the big source of truth but the evals do show some improvements (5-10%) with these directions. We should get a more detailed report about this from the nv-skills report.

@danielfrg

Copy link
Copy Markdown
Contributor Author

/nvskills-ci

Signed-off-by: nvskills-svc-account <svc-nvskills-signing@nvidia.com>

@coderabbitai coderabbitai Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actionable comments posted: 1

Nitpick comments: 1

skills/cuda-python/BENCHMARK.md: 1

64-64: Trivial · Maintainability & Code Quality · Quick win

suggestion: Identify where the detailed report lives.

Line 64 tells readers to review the detailed report before they rely on a partial dimension. The file does not say where that report is, and it is not in the checkout. Line 147 and skill-card.md:89 have the same gap. Add a durable location for the report, or state that it is held outside the repository and how to request it. Apply the same change to the two other references.

Review info

Run configuration

Commits

Reviewing files that changed from the base of the PR and between f53f3d9 and 01d7742.

Files selected for processing: 3

  • skills/cuda-python/BENCHMARK.md
  • skills/cuda-python/skill-card.md
  • skills/cuda-python/skill.oms.sig

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 11 remain after this review.

@@ -0,0 +1 @@
{"mediaType":"application/vnd.dev.sigstore.bundle.v0.3+json","verificationMaterial":{"x509CertificateChain":{"certificates":[{"rawBytes":"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"},{"rawBytes":"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"},{"rawBytes":"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"}]},"tlogEntries":[]},"dsseEnvelope":{"payload":"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","payloadType":"application/vnd.in-toto+json","signatures":[{"sig":"MGUCMQC395i6+DTmGbsIjm07TRmIecMc6EVChJ7QjGwgfX6gMlylf4si5VuZ7F7IacXwZpsCMHHzgAyD6rXZiaDxMC9YdDcfRwdskNq6Wdw31657ZGaOYpzzuQ+R3Gy4N067grnPWg==","keyid":""}]}} No newline at end of file

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Major · Functional Correctness · Quick win

important: Exclude the signature from SPDX header validation.

The check-spdx hook includes this .sig file. .spdx-ignore excludes *.json and *.md, but not *.sig. This file has neither required SPDX field, so the checker returns failure. Its --fix mode does not add missing fields. (raw.githubusercontent.com)

Add an exact-path exemption for skills/cuda-python/skill.oms.sig. Do not insert headers into the signature bundle.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@danielfrg
danielfrg requested a review from mdboom November 17, 2026 15:23
@danielfrg

Copy link
Copy Markdown
Contributor Author

@mdboom can you do another quick pass here, nothing changed other than the required CI for nvidia/skills.

@danielfrg

Copy link
Copy Markdown
Contributor Author

/ok to test c695018

@danielfrg

Copy link
Copy Markdown
Contributor Author

@rwgk do you know why these pathfinder CI are failing?

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cuda.bindings Everything related to the cuda.bindings module cuda.core Everything related to the cuda.core module cuda.pathfinder Everything related to the cuda.pathfinder module documentation Improvements or additions to documentation P1 Medium priority - Should do

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants