UX: Clarify what happens after a failed test under Testing
Resolved 💬 3 comments Opened Oct 28, 2025 by fedepalu2 Closed Jan 17, 2026
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
Type: documentation
Summary
Under Testing, when a test fails it's unclear whether the failure will be fixed automatically or documented.
Request
Right after the failure info, show whether the system will attempt a fix, open a follow-up, or document the limitation.
3 Comments
Can you provide more context? Are you talking about failures for tests that are run by the codex agent? Which version of codex are you using: the web version, the CLI?
I'm on the web, talking about any answer. In the answer there is "Test" paragraph but is not clear how it works, it is some how automated, some how I can ask test, somehow is the code has some test it may be reported. Depends... on the agent but sometimes is not clear whether I can orient a test, and if a test fails and the agent retried and solved the problem. Finally the point is that this paragraph could be documented in the behaviour
I'm on the web, talking about any answer. In the answer there is "Test" paragraph but is not clear how it works, it is some how automated, some how I can ask test, somehow is the code has some test it may be reported. Depends... on the agent but sometimes is not clear whether I can orient a test, and if a test fails and the agent retried and solved the problem. Finally the point is that this paragraph could be documented in the behaviour