codex-chats-mcp
Manages ChatGPT conversations and Codex Cloud tasks through MCP, enabling list, search, rename, archive, export, and permanent deletion from any MCP-compatible client.
README
codex-chats-mcp
An MCP server for managing ChatGPT conversations and Codex Cloud tasks (chats) from any MCP-compatible client — Claude Code, Codex, Cline, etc.
ChatGPT's web UI lets you archive chats but the "Delete all" button only wipes visible ones, and Codex Cloud has no per-task delete at all. This server wraps the internal chatgpt.com/backend-api endpoints so you can list, search, rename, archive, export, and permanently delete both kinds of chats from your agent.
Unofficial. This uses undocumented internal endpoints (
/conversations/*,/wham/tasks/*). They can change without notice and require a valid ChatGPT session. Use at your own risk.
Install
pip install codex-chats-mcp
Or with uv:
uv tool install codex-chats-mcp
This installs a codex-chats-mcp executable.
Authentication
The server reads ~/.codex/auth.json — the same file the Codex CLI maintains after codex login. If you don't have Codex installed, log in once with npx @openai/codex login (or sign in via the Codex desktop app) to produce the file.
The token has full access to your ChatGPT account. Treat the auth file as a secret.
Wire it up
Codex CLI (~/.codex/config.toml)
[mcp_servers.codex-chats]
command = "codex-chats-mcp"
Claude Code
claude mcp add codex-chats codex-chats-mcp
Anything else
Point your MCP client at the codex-chats-mcp executable. It speaks MCP over stdio.
Tools
ChatGPT conversations (the "Recents" list)
| Tool | What it does |
|---|---|
list_conversations |
Paginates through your conversations. Filters out archived by default. |
get_conversation |
Full payload for one conversation, including the message tree. |
search_conversations |
Substring match on titles (client-side). |
rename_conversation |
Change a conversation's title. |
archive_conversation / unarchive_conversation |
Toggle the archive flag. |
delete_conversation |
Permanently delete one chat (is_visible=false). No undo. |
delete_conversations_matching |
Delete every chat whose title matches a substring. Requires confirm=True. |
delete_all_conversations |
Nuke every visible chat — same as ChatGPT's "Delete all chats" button. Requires confirm=True. |
export_conversations |
Dump titles/IDs (and optionally full message trees) to a JSON file. |
Codex Cloud tasks
| Tool | What it does |
|---|---|
list_chats |
Paginate Codex tasks, filterable by all / current / archived. |
get_chat |
Summary of one task. |
get_chat_raw |
Full raw task payload. |
archive_chat / unarchive_chat |
Toggle archive state. |
delete_chat |
Permanently delete a task. Works on active OR archived. |
delete_all_archived |
Bulk-delete every archived task. Requires confirm=True. |
Safety
Every destructive bulk action (delete_all_conversations, delete_all_archived, delete_conversations_matching) requires confirm=True. Without it the tool returns a preview of what would be deleted. There is no undo on the ChatGPT side.
Development
git clone https://github.com/shoyu-ramen/codex-chats-mcp
cd codex-chats-mcp
python3 codex_chats_mcp.py
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.