Codex App renders ChatGPT-generated Markdown download links as non-clickable text
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 52/100
Research direction
Start by reproducing the generated Markdown attachment flow in the Codex App on macOS and compare it with the ChatGPT web client, focusing on message rendering and sandbox:/mnt/data/ references. Done means generated files render as interactive attachments and can be clicked and downloaded when the conversation is opened in the app.
Written by the indexing model from the issue text.
Description
What version of the Codex App are you using (From “About Codex” dialog)?
26.721.81911 (5973)
What subscription do you have?
Not provided
What platform is your computer?
macOS 26.5.2 (Build 25F84), Darwin 25.5.0, arm64
What issue are you seeing?
A Markdown file generated in a ChatGPT conversation is downloadable in the ChatGPT web client, but the same conversation does not expose a working file download in the Codex App.
In the web client, the generated .md file is rendered as an interactive file attachment/download component. In the Codex App, the filename or download text is rendered as ordinary text with no link styling and no click behavior. A raw fallback reference such as:
sandbox:/mnt/data/authentik-docker-to-kubernetes-migration-runbook.md
is not resolved into a clickable attachment or download target by the Codex App.
This appears to be a client-side message/attachment rendering regression rather than a file-generation or storage failure, because the same file remains accessible from the web client.
This affects Markdown documents as well as potentially other generated deliverables such as configuration files, scripts, reports, and architecture documents.
Related broader report: #34832 (Issue 2). This report is intentionally focused on the cross-client rendering mismatch and the sandbox:/mnt/data/ attachment reference behavior.
What steps can reproduce the bug?
- In ChatGPT, ask the model to generate a downloadable Markdown file.
- Wait for the response containing the generated file attachment/download link.
- Open the conversation in the ChatGPT web client.
- Confirm that the file is displayed as an interactive attachment and can be downloaded.
- Open the same conversation in the Codex App.
- Observe that the generated filename or
sandbox:/mnt/data/...reference is rendered as plain text. - Click the displayed filename/text.
- Nothing happens; the file cannot be opened or downloaded from the Codex App.
What is the expected behavior?
Codex App should match the web client behavior:
- Render generated files as interactive attachment/download components.
- Resolve
sandbox:/mnt/data/...references used by ChatGPT-generated files. - Allow users to click and download the generated file directly.
- Preserve access to file attachments when a conversation is opened across clients.
Additional information
Two comparison screenshots are available and will be attached: one showing the working interactive file attachment in the web client, and one showing the non-clickable plain-text rendering in the Codex App.
Observed on 2026-07-29 (Asia/Shanghai).
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.5k
- Avg merge
- 1m
- Merged PRs (30d)
- 1k
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from openai/codex
-
enhancement remote
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
-
bug CLI windows-os
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
-
macOS sandbox blocks hw.optional.arm64 sysctl, causing Flutter to misdetect Apple Silicon as x64 Openbug CLI sandbox
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
-
bug CLI TUI
Difficulty 2/5 1-3 hours Newbie friendliness 90/100
-
CLI config enhancement skills
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 86/100
kwakseongjae/auto-hwp#319 ·
-
area:cli bug filter-quality good first issue priority:medium
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
-
Difficulty 1/5 Under an hour Newbie friendliness 72/100
bevyengine/bevy#25861 ·
-
comp-datalake
Difficulty 2/5 1-3 hours Newbie friendliness 88/100
ClickHouse/ClickHouse#121222 ·
-
A-linter
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
oxc-project/oxc#26863 ·