Wish: Native MCP Server in ChatGPT
RFOK · reddit · 2026-07-13
This Reddit post proposes adding a native MCP server to Codex and ChatGPT desktop, not just a client.
The core idea is to allow other authorized apps to use Codex/ChatGPT as a callable agent: local IDEs, automation tools, or multi-agent orchestrators can start tasks, pass context, monitor progress, receive structured results, and cancel if needed.
The author emphasizes a strict permission model must accompany this:
- Off by default, ideally localhost only
- Explicit capabilities and workspace scope
- User confirmation for destructive or external operations
- Visible session state and revocable access
- Audit logs per request
- Maintain existing plan limits, billing, and security policies
The author acknowledges that open automation may conflict with OpenAI's business model, but argues MCP could become a universal interface for 'tool-calling agents', provided users always control permissions.
More from coding & agent
- A roundup of AI agents and MCP resources, including how to evaluate agents — _jaydeepkarale · 2026-07-21
- A full course shows how to build and deploy an AI agent with OpenAI and LangChain — _jaydeepkarale · 2026-07-21
- A beginner guide to AI agents points readers to a Stanford webinar — _jaydeepkarale · 2026-07-21
- A practical guide on how to evaluate AI agents — _jaydeepkarale · 2026-07-21
- MCP is headed toward easier scale, event-driven extensions, and workable file uploads — EricBuess · 2026-07-21
- Developers debate the missing composition model for AI agents — threepointone · 2026-07-21