Codex plugin: choose which past images go to the model with your next message. Unchecked images become placeholders, the task goes on, and requests stay small.
Choose which past images go to the model with your next message, without leaving the task.
I was iterating on image generation with GPT in Codex. After a dozen or so images, the task started failing: the connection kept dropping, and the retries every few minutes never went through. Digging into it, I found that Codex sends every image in the task's history again with every request, and each request had grown to about 44 MB. The only way out was a new task, which meant losing the conversation.
So I wanted to keep the long conversation going and decide, turn by turn, which images the model still gets. Uncheck the ones you no longer need: from the next message on they become short placeholders, and the task and the conversation stay the same.
- Every image in one list: a Context Assets tab in Codex's side panel lists every image in the task, turn by turn: the ones you uploaded and the ones the model viewed or generated, with thumbnails, IDs and sizes. Click one to preview it. - Filter by source: with many images, show only some sources: your uploads, PDF pages you commented on in Codex (with the page number and the PDF's name), web page screenshots, and images the model viewed, got from a tool or generated. Filtering changes what you see, never what is sent. - One click to leave an image out: unchecked images become a short placeholder from the next message on. Toggle single images or a whole turn. - Duplicates handled: when an identical image is still being sent, the placeholder points to it. - The model knows why: a note tells the model the images were left out by you to
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/chipfighter--codex-attachment-manager.html)
LayerMap MCP: a layered map of your codebase for coding agents: callers, callees and the impact of a change, for TypeScript, JavaScript, Go, Python and Java. MCP server, CLI, and plugins for Claude Co
wakeup595626-cmyk/mcphelmOpen-source control center for the OpenAI Secure MCP Tunnel: register local MCP servers (stdio/HTTP), run and supervise the official tunnel-client, and connect ChatGPT / Codex through a desktop app, l
adityajha2005/decisionkitFast, cheap, probability-scored decisions for coding agents, scripts and CI. CLI + MCP server + Claude Code plugin for the OpenAI Decisions API and Jev (TypeSafe AI).
Araz0-0dev/am-codeAM Code — agentic coding assistant (VS Code + Windows/Linux/macOS desktop) with your own models, MCP support and a token saver that compresses every request. | دستیار کدنویسی ایجنت با مدلهای خودت
gonzaroman/IHMT-MEMORYLong-term memory for AI coding agents (Claude Code, Codex, opencode) as a tree of plain files
Vuk97/codex-remote-proChatGPT <-> Codex bridge: local coding agents ask ChatGPT for advice, and ChatGPT can inspect or steer your running Codex sessions. No API key.
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.