regen-snapshots

Regenerate snapshot JSON files for CPU, SYCL, and Netflix benchmarks.

3|1|Updated May 28, 2026
One-click install
npx skills add https://github.com/VMAFx/vmafx --skill regen-snapshots
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: regen-snapshots
Source: https://github.com/VMAFx/vmafx/tree/main/.claude/skills/regen-snapshots
Command: npx skills add https://github.com/VMAFx/vmafx --skill regen-snapshots

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill keeps fork-added test snapshots in sync after intentional numerical changes, so maintainers can refresh golden JSON outputs without doing the regeneration workflow by hand.

Core Features & Use Cases

  • Snapshot Regeneration: Rebuilds the requested backend and regenerates matching CPU and Netflix benchmark snapshot files.
  • Validation and Diffing: Compares old and new outputs, skips identical files, and records affected frames and maximum delta.
  • Controlled Updates: Enforces justification, protects unrelated test areas, and drafts a commit message instead of auto-committing.
  • Use Case: After a model, backend, or numerical pipeline change, use this Skill to refresh only the affected snapshot files with traceable reasoning.

Quick Start

Ask the Skill to regenerate the affected snapshot files for your intentional change and include a short justification plus the backend scope.

Frequently Asked Questions about regen-snapshots

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I regenerate snapshot JSON files after intentional numerical changes?

To regenerate snapshot JSON files after intentional numerical changes, provide a non-empty justification and backend scope to trigger backend-specific rebuilds, tempfile-based regeneration, diff comparison, and guarded replacement of golden outputs.

What is the safest way to refresh golden test outputs for CPU and Netflix benchmarks?

The safest way to refresh golden test outputs for CPU and Netflix benchmarks is using a tempfile-based regeneration process that compares old and new outputs, skips identical files, and only replaces validated deltas that stay within defined limits.

Do I need to provide a justification when updating regression testing snapshots?

Yes, you need to provide a non-empty justification when updating regression testing snapshots because the process enforces controlled updates and drafts a commit message with traceable reasoning instead of auto-committing.

What's the best way to handle snapshot maintenance for SYCL backend benchmark results?

The best way to handle snapshot maintenance for SYCL backend benchmark results is requesting backend-specific rebuilds that protect unrelated test areas while refreshing only the affected snapshot files with diff comparison.

Why does snapshot regeneration skip identical JSON files?

Snapshot regeneration skips identical JSON files because the validation and diffing mechanism compares old and new outputs to avoid unnecessary updates, recording only affected frames and the maximum delta for changed files.

Can I auto-commit refreshed benchmark snapshot files directly?

No, you cannot auto-commit refreshed benchmark snapshot files because the process drafts a commit message instead of auto-committing, ensuring controlled updates with traceable reasoning for intentional numerical changes.