quark-torch-debug
Diagnose failed Quark Torch PTQ installation, execution, script generation, or export attempts. Use when the user reports a torch-side error, stack trace,…
Skills
Find an Agent Skill for your task. Read the guide and copy the install command.
Updated 10-31 00:15
Testing & quality
12,987 results
quark-torch-llm-eval
End-to-end LLM accuracy evaluation on AMD ROCm (ROCm-only) — docker container setup OR host (no-docker) runtime, vLLM/SGLang/ATOM serving, lm-eval / lighteval…
code-review
Principal-engineer review of a GitHub PR via two-phase approval. Pre-checks PR description quality, size, and single-concern scope. Generates findings across…
creating-webhooks
Creates durable webhook handlers in Amp plugins running in Orbs. Use when asked to create, configure, or test a plugin webhook.
publish-gist-report
Publish E2E/QA test reports (markdown + screenshots) as secret GitHub Gists. Uses two-gist pattern to work around GitHub rendering limits. Trigger when: report…
gm-eliminate
Third stage of resolving one PRD row: eliminates each unknown by a live JIT run, adding new unknowns found on the way. Loads gm-edit only when none is open.…
gm-validate
Verification lane for one PRD row, spawned by the orchestrator after an executor hands off: validates the pushed edit with live JIT runs as a witness in its…
review-changes
Review uncommitted code changes for bugs, security vulnerabilities, and quality issues. Use before commits or when asked to review code.
tests-developer
Smart router to testing patterns and practices. Use when writing unit tests, creating mocks, testing edge cases, or working with Swift Testing and XCTest…
run-e2e
Run SkyWalking E2E tests locally
minervia-update
Update Minervia to the latest version while preserving your customizations. Creates backups, shows what's new, lets you choose how to handle conflicts. Use…
verification-before-completion
Use when about to claim work is complete, fixed, or passing. Requires running verification commands and confirming output before making any success claims.…
gps-testers
Manage Google Play track testers (email lists / Google Groups) with gps. Use when adding closed testing users, updating internal/alpha/beta testers, or reading…
shots
Use when an agent should capture, publish or read screenshots and movies with `archdev shots` — checking whether Shots is on, publishing stills of UI work into…
test-coverage
Finds complex business logic, edge cases, corner cases, and past regressions that lack tests, then writes the missing unit tests (and integration/E2E tests…
genai-conformance
Run, interpret, and iterate on the OpenInference GenAI conformance MVP at python/openinference-instrumentation/scripts/conformance/. Use when the user mentions…
deslop
Use when the user asks to deslop, clean up, de-slop, harden, simplify, deduplicate, or improve code quality across an existing codebase, especially when the…
writing-tests
Read BEFORE writing, adding, changing or reviewing any test (unit, integration, component, e2e), including tests added while implementing a feature or fixing a…
external-cannbot-model-model-infer-runtime-debug
Skill for diagnosing and fixing runtime errors in Ascend NPU model inference based on the PyTorch framework. Systematically troubleshoot runtime errors across the entire链路 of model loading, initialization, and inference execution, including aicore timeout, HCCL communication errors, OOM, operator constraint violations, inference hangs, etc. Trigger scenarios: NPU runtime errors (RuntimeError, aicore timeout…
external-cannbot-ops-pypto-precision-compare
PyPTO operator precision issue debugging skill. Provides two precision comparison methods: the file save method (using pypto.pass_verify_save and torch.save) and the binary comparison method (using checkpoint tensors). Use this skill when you need to debug PyPTO operator precision, locate the source of precision differences, or perform intermediate result comparison.
external-gitcode-ascend-atb-csv-tester
Run ATB (Ascend Transformer Boost) CSV tests. Invoke this skill when the user needs to execute CSV-format ATB test cases, verify operator correctness, or run any CSV test file under ATB. It must be used with the CANN environment and a compiled ATB test framework.
inference-precision-tensor-dump-compare
Model-layer Tensor instrumentation and precision comparison tool. Used to capture intermediate tensors of each model layer during the model forward pass, enabling GPU/NPU precision comparison and debugging. Supports the vLLM and SGLang inference frameworks. When to use: When you need to debug precision issues between GPU…
gitcode-pr-audit
Quality audit for merged GitCode PRs: sample by time range or repo list, check compliance (labels, comments, tests, size, etc.), output table. Use when user…
ab-test-setup
When the user wants to plan, design, or implement an A/B test or experiment. Also use when the user mentions "A/B test," "split test," "experiment," "test this…