ProDex Chat Model icon

ProDex Chat Model

Use Codex with your ChatGPT subscription as the chat model behind n8n AI Agent and other LangChain nodes.

ProDex Chat Model

AI-generated

Summary

Supplies a Codex chat model for n8n AI Agent and other LangChain nodes. It converts incoming messages to a text prompt, adds the optional system prompt and installed skills, then returns Codex’s response as assistant text; it does not provide native tool-call payloads.

Inputs

  • Use n8n Credentials — Off by default: uses authentication saved by ProDex Setup. If enabled, select a ProDex Auth API credential.
  • System Prompt — Static instructions prepended to the prompt sent to Codex.
  • Skills — Select installed skills to include with each request. Install skills with ProDex → Install Skill first.
  • Model — Codex model to use: GPT-5.5, GPT-5.4, or GPT-5.4 Mini. Defaults to GPT-5.4.
  • Reasoning Effort — Codex reasoning level: Low, Medium, High, or Extra High. Defaults to Medium.
  • Personality — Codex personality: Default, Friendly, or Pragmatic. Defaults to Default.
  • Sandbox — Filesystem access for Codex: Read Only, Workspace Write, or Full Access. Defaults to Read Only.
  • Timeout (Seconds) — Maximum time for a Codex run, in seconds; defaults to 300, with a minimum of 30.

Output shape

An AI language model connection that produces assistant text responses.

Connect the Model output to an AI Agent Chat Model input. First run ProDex Setup → Start Device Login, complete browser authentication, then run Wait for Login Complete. The model starts a new Codex thread for each generation. AI Agent tool nodes are not fully supported: tool-related messages are rendered into the text prompt, and Codex returns text rather than native tool-call payloads.

Links

Discussion