wings-engine-patch

Patch AI inference engines at runtime via wrapt import hooks.

3|1|Updated Feb 2, 2026
One-click install
npx skills add https://github.com/sunchendd/good_skills --skill wings-engine-patch
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: wings-engine-patch
Source: https://github.com/sunchendd/good_skills/tree/main/monkey-patch
Command: npx skills add https://github.com/sunchendd/good_skills --skill wings-engine-patch

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Developers need a safe, non-intrusive way to modify runtime behavior of AI inference engines without touching source code.

Core Features & Use Cases

  • Non-intrusive runtime patching using Python import hooks (wrapt)
  • Feature-based patch management with propagation and deduplication
  • Version-scoped patches that target specific engine versions
  • Clear guidance for safe patching, testing, and rollback in production environments

Quick Start

Set the WINGS_ENGINE_PATCH_OPTIONS environment variable to configure patches and run Python to apply them.

Frequently Asked Questions about wings-engine-patch

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

FAQPage Schema
How do I apply non-intrusive runtime patches to an AI inference engine without modifying source code?

Non-intrusive runtime patching uses Python import hooks via wrapt to modify AI inference engines like vllm without changing original packages, applying feature-based changes through a centralized registry.

What is the best way to manage version-scoped patches for vllm and vllm_ascend?

Version-scoped patching targets specific engine versions using a centralized registry that handles patch propagation and deduplication, ensuring only compatible changes apply to your vllm or vllm_ascend environment.

How do I configure and activate runtime patches using the WINGS_ENGINE_PATCH_OPTIONS environment variable?

To configure runtime patches, set the WINGS_ENGINE_PATCH_OPTIONS environment variable with desired patch options, then run Python; a .pth file triggers the startup mechanism to apply them automatically.

Does non-intrusive runtime patching work without installing new dependencies beyond wrapt?

Non-intrusive runtime patching requires the wrapt library for Python import hooks and a .pth file for the startup mechanism, needing no other dependencies to patch AI inference engines.

Can I safely test and roll back runtime patches in production AI inference environments?

Runtime patching provides clear guidance for safe patching, testing, and rollback in production environments, using feature-based management and a patch registry to control changes without modifying source code.