review-op

Review a CV-CUDA operator end-to-end (support / test / bench / docs coverage). Use when the user asks to review an operator, audit its input-type/layout/dtype support, test coverage, benchmark coverage, or docs/API, or to find & fix per-operator coverage gaps. Produces a deterministic findings-first report.

Jump to install Review files

Repository: CVCUDA/CV-CUDA
Last source activity: August 11, 2026 at 21:59
Detected SKILL.md language: English
Stars: 2,729
Forks: 264

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Install with an AI assistant

Copy this prompt into the AI assistant you're using.

Follow https://skillsmp.com/skill-install/prompt.md to install only the Agent Skill at https://github.com/CVCUDA/CV-CUDA/tree/main/.agents/skills/review-op.

A direct command skips the review prompt. Inspect the source before running it.

npx skills add https://github.com/CVCUDA/CV-CUDA --skill review-op

The command stays on one line. Scroll horizontally to inspect it before copying.

Prefer a local copy? Download the files currently available to SkillsMP.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

SKILL.md

Source instructions · Read-only preview

name: review-op
description: Review a CV-CUDA operator end-to-end (support / test / bench / docs coverage). Use when the user asks to review an operator, audit its input-type/layout/dtype support, test coverage, benchmark coverage, or docs/API, or to find & fix per-operator coverage gaps. Produces a deterministic findings-first report.

[//]: # "SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved." [//]: # "SPDX-License-Identifier: Apache-2.0" # Review Op Thin entry point. The checklist substance and the checks are shared repo artifacts: the deterministic checker `tools/review_op.py` and the spec `.agents/guidance/REVIEW_OP_GUIDELINES.md`. Keep this skill thin; do not duplicate the checklist here. ## Workflow 1. Run `python3 tools/review_op.py <Operator>` (scope with `--domain support|test|bench|docs`; `--format json` for machine output; `--run` for the GPU-run-dependent bench checks). The checker is read-only and deterministic. 2. Interpret each finding by its cited item id in `.agents/guidance/REVIEW_OP_GUIDELINES.md`. Resolve every `MANUAL` item by inspecting the location it points to. 3. If the user asked to fix (`--fix`), apply the corrective action named per `GAP` in `.agents/guidance/REVIEW_OP_GUIDELINES.md`, then **re-run the checker to prove the item is `PASS`**. Follow the inviolable `--fix` rules in `.agents/guidance/REVIEW_OP_GUIDELINES.md` ("### `--fix`"). 4. Return findings-first: `GAP`s, then unresolved `MANUAL`s, then `RECOMMENDATION`s, then the per-domain + overall verdict. Completion = a re-run shows zero GAP and zero unresolved MANUAL. ## Prompt Handling Requests like "review the Resize operator", "/review-op CenterCrop", "audit Flip's bench coverage", or "does Remap cover every declared layout?" trigger this skill. Use `--domain` to scope to a single domain when the request is domain-specific.

View on GitHub

Back to search Back to installation

make-op

Add a new CV-CUDA operator end-to-end per .agents/guidance/MAKE_OP_GUIDELINES.md, with a deterministically-enforced definition-of-done. Use when asked to create/add a new operator, scaffold one, or verify that a new operator is complete (approved spec, wired scaffold, bit-exact regression tests across the declared support matrix, required layout parity, complement negatives, docs + relnote, benched dtypes, tests that run and pass, optimization-ready).

2026-09-16

refactor-op

Find and safely apply per-operator refactoring / redundancy-reduction opportunities in a CV-CUDA operator (near-duplicate Tensor/VarShape kernels, reinvented shared utilities, dead code). Use when asked to reduce code duplication, de-duplicate or unify an operator's kernels/bindings, remove dead code, or verify a refactor changed nothing observable. Produces a deterministic findings-first report; the apply path is gated by a strict bit-exact / feature-set / test-coverage parity check.

2026-09-16

make-op-scaffold

Scaffold a new CV-CUDA operator — a complete, wired, building skeleton — and delegate the implementation to a human or another AI. Use when asked to set up / stub out a new operator without implementing it, or to bootstrap one for someone else to finish. Supports --bare (skip the spec when the definition itself is delegated).

2026-08-11

View all from this repository

review-op

Source facts

Install options

Install with an AI assistant

Review the source files

More from this repository

Install with an AI assistant

More from this repository