codex-custom-provider-pack

Name: codex-custom-provider-pack
Availability: InStock
Author: KwongFuk

Configure Codex for custom provider endpoints and maintain model registries.

Updated Apr 10, 2026

Search Tags:#automation #codex #jsonl #smoke-test #model-registry #custom-provider #cache-scan

Ready to run

npx skills add https://github.com/KwongFuk/codex-skills --skill codex-custom-provider-pack

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: codex-custom-provider-pack
Source: https://github.com/KwongFuk/codex-skills/tree/main/workspace/codex-custom-provider-pack
Command: npx skills add https://github.com/KwongFuk/codex-skills --skill codex-custom-provider-pack

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Configure Codex to operate against a custom provider endpoint and maintain a registry of hosted and local models, enabling consistent access and evaluation across environments.

Core Features & Use Cases

Configure Codex CLI, desktop, or editor plugins to point at the custom provider at https://api.mol.us.ci/v1.
Maintain a live table of hosted models and a local cache registry to streamline model discovery and switching.
Use the included helper scripts to scan Hugging Face caches under /scratch/gguo/.hf_cache/hub, smoke-test API access, and generate JSONL samples for evaluation.

Quick Start

Install and run the helper workflow to validate configuration and generate sample data with the provided scripts.

Frequently Asked Questions about codex-custom-provider-pack

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

FAQPage Schema

How do I configure Codex to use a custom provider endpoint?▼

To configure Codex with a custom provider, point your CLI, desktop, or editor plugins to the desired API endpoint. This Skill sets the provider URL and maintains a registry of hosted and local models for consistent access.

Can I scan my local Hugging Face cache to register models for Codex?▼

Yes, you can scan local Hugging Face caches to populate a model registry. The included helper scripts enumerate models via the /models endpoint and scan cache directories to streamline local model discovery.

Does this custom provider setup store my API credentials in the code?▼

No, configuring the custom provider avoids credential storage in code. The Python helper script manages API access and data generation without embedding credentials directly in the source files.

What is the best way to generate JSONL evaluation samples for custom models?▼

The best way to generate JSONL evaluation samples is using the provided Python helper script. It automates data generation for evaluation while interacting with the custom provider endpoint.

How do I smoke-test API access for a custom model provider?▼

You smoke-test API access by running the included helper workflow. The scripts validate the configuration against the custom provider endpoint and ensure the API connection is functioning correctly.