add-codex

Configure OpenAI Codex as a NanoClaw agent provider with flexible authentication.

5|Updated Mar 14, 2026
One-click install
npx skills add https://github.com/chiptoe-svg/nanoclaw_gccourse --skill add-codex-chiptoe-svg
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: add-codex
Source: https://github.com/chiptoe-svg/nanoclaw_gccourse/tree/main/.claude/skills/add-codex
Command: npx skills add https://github.com/chiptoe-svg/nanoclaw_gccourse --skill add-codex-chiptoe-svg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

NanoClaw defaults to the Anthropic-only Claude Agent SDK for agent operations, creating vendor lock-in for teams that prefer OpenAI's Codex CLI and ecosystem. This Skill eliminates that limitation by enabling full Codex integration as a drop-in agent provider.

Core Features & Use Cases

  • End-to-End Codex Setup: Automatically fetches Codex provider source files from the upstream providers branch, wires them into host and container codebases, and updates the Dockerfile to install the Codex CLI with no manual file edits required.
  • Flexible Authentication Support: Works with ChatGPT subscriptions, OpenAI API keys, and custom OpenAI-compatible endpoints (including local LLM servers via Groq, Together, or vLLM) to match your existing workflow.
  • Full Feature Parity: Unlocks Codex's native session resume, streaming events, MCP tool access, and built-in context management, matching the Claude Agent SDK's feature set without Anthropic dependency.
  • Use Case: Development teams using OpenAI tools can run NanoClaw agents with Codex instead of Claude, leveraging existing OpenAI credentials and tooling without rewriting agent logic.

Quick Start

Use the add-codex skill to install the Codex agent provider for your NanoClaw instance, configure your preferred authentication method, and rebuild the container to start using Codex-powered agents.

Frequently Asked Questions about add-codex

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

FAQPage Schema
How do I use OpenAI Codex as an agent provider in NanoClaw?

To use Codex as an agent provider, NanoClaw fetches upstream provider source files, wires host and container codebases, and updates the Dockerfile to install the Codex CLI automatically. This replaces the default Anthropic Claude Agent SDK without manual file edits.

Can I connect NanoClaw to a local LLM server instead of OpenAI?

Yes, NanoClaw supports flexible authentication for custom OpenAI-compatible endpoints. You can configure the Codex provider to route agent operations through local LLM servers using Groq, Together, or vLLM instead of relying on a standard OpenAI API key.

Does the Codex integration for NanoClaw support MCP tools and session resume?

Yes, the Codex integration provides full feature parity with the Claude Agent SDK. It unlocks Codex's native session resume, streaming events, MCP tool access, and built-in context management for agent workflows.

How do I set up Docker for a Codex-powered NanoClaw deployment?

Setting up Docker requires applying the Codex provider files and rebuilding the container. The process automatically updates the Dockerfile to install the Codex CLI and configures host wiring for your chosen authentication method.

What is the best way to eliminate Anthropic vendor lock-in for NanoClaw agents?

The best way to eliminate vendor lock-in is adding the Codex provider. It enables full Codex integration as a drop-in agent provider, allowing development teams to leverage existing OpenAI credentials and tooling without rewriting agent logic.

Do I need a ChatGPT subscription to run Codex agents in NanoClaw?

No, a ChatGPT subscription is not strictly required. The authentication configuration supports ChatGPT subscriptions, standard OpenAI API keys, and custom OpenAI-compatible endpoints to match your existing workflow.