Codex should proactively propose transport/workaround alternatives when the user's goal is clear
Feedback
Codex should more proactively propose and execute viable workaround paths when the user's desired outcome is clear but the first transport or integration path fails.
User-observed case
A user needed to send a large, full-resolution Word document. The initial obvious paths failed:
- email attachment exceeded the mail server attachment limit
- a connected Drive upload path had a lower upload limit than the file size
Codex initially treated those failures too narrowly. The better behavior would have been to infer that the user's real goal was still achievable and then offer concrete alternatives without waiting for the user to spell them out.
Desired behavior
When Codex hits an operational blocker such as an attachment limit, connector upload cap, file-size limit, authentication gap, or unavailable tool, it should:
- keep the user's underlying goal active rather than stopping at the failed path;
- identify plausible workaround classes itself, such as cloud links, extracting the target file from an archive, compression tradeoffs, alternative upload providers, local sync folders, or asking for narrowly scoped consent for a third-party upload;
- distinguish between preserving the exact artifact and producing a smaller substitute;
- ask for consent when the workaround changes privacy, destination, fidelity, or retention characteristics;
- once consent is given, execute the workaround end to end and update the user-facing deliverable.
Why it matters
This is a common agentic reliability issue: the user often knows the destination, not the transport mechanics. Codex should act as the problem-solving layer and surface the best next transport path, rather than requiring the user to propose the workaround.
Component
Codex app / agent behavior / tool-use planning.
Privacy note
This feedback intentionally omits private file names, links, recipients, and business context. The point is the general behavior around failed transport paths and proactive alternatives.