openai / openai/codex

Remote compaction repeatedly fails with content_filter on literary material

Open
#43,969 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug context model-behavior
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

What happened

Codex repeatedly fails while compacting a conversation involving literary material:

Error running remote compact task: stream disconnected before completion: Incomplete response returned, reason: content_filter

The task was to process existing literary text, not to request harmful instructions. This looks like a false positive, but I do not know which passage or category triggered it.

The immediate problem is that an internal compaction failure prevents the session from continuing. Retrying has repeatedly produced the same error and cost time and usage without recovering the session.

Environment

Windows, Codex desktop app. The app version was not recorded at the time of the original report.

Reproduction and diagnostics

This has happened in multiple sessions. I do not yet have a minimal reproduction that reliably triggers it. The affected session was submitted through the in-app feedback dialog with diagnostic logs:

Feedback ID: 01a07f8e-bc37-70a2-961f-8a5420690234

The comments below contain identifiers for subsequent occurrences. I am keeping the source text and full conversation out of this public report for privacy.

Expected behavior

Quoted literary content should be evaluated in the context of the task. If compaction cannot complete, Codex should provide a clear explanation and a recovery option rather than repeatedly retrying and leaving the session unable to continue.

Please investigate both the apparent false positive and the handling of content_filter during remote compaction.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the diagnostic logs for feedback ID 01a07f8e-bc37-70a2-961f-8a5420690234 and the remote compaction error path. Compare the logged occurrences to determine whether the content_filter response is reproducible, then verify that an unsuccessful compaction reports a recovery option instead of repeatedly retrying and blocking the session.

Written by the indexing model from the issue text.

Assessment

Domain
ai-infra-agents, backend-api-design
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.