← sources

How do you get Codex, Claude Code and a planner agent working together in one place instead of copy-pasting between them?

source: reddit (reddit:AI_Agents)published 2026-11-14 15:03ingested 2026-11-15 02:33

retained text

I'm building an AI operations system for a small business.



My current setup works, but it's all manual:



\- Builder: Codex (GPT)

\- Planner: Claude Opus in a chat (decisions, scope, writing the prompts)

\- Reviewer: Claude Code CLI checks every change before it goes live

\- Sometimes a second model does an extra review



The problem is I'm the messenger. I copy the planner's prompt into Codex, copy Codex's reply to the reviewer, then copy the review back to the planner. It's slow, and it falls apart when I'm busy.



What I want:



\- All the agents in one place, passing work to each other

\- Builder and reviewer from different AI companies, so they don't share blind spots

\- A reviewer and a security checker running on every change

\- A "scout" agent that keeps an eye out for new tools, repos, model releases and security issues that could improve the build

\- Me approving anything that goes live. That part's non-negotiable



I'm on Windows, I already pay for ChatGPT (Codex) and Claude, and the project lives in a local git repo (private client data, so no public GitHub).



What are you actually using for this? Real-world experience, good or bad, would help a lot, especially on keeping it safe when agents talk to each other without me in the middle.

document fetched 2026-11-15 00:18 · canonical https://reddit.com/r/AI_Agents/comments/1xw4j71/how_do_you_get_codex_claude_code_and_a_planner