Repository navigation
fix(sandbox): preserve UTF-8 boundaries in Cloudflare SSE - #4707
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 54f9430aaf
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
All prior review feedback has been addressed in the current head ( |
|
Codex Review: Didn't find any major issues. 👍 Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
Security review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
Summary
Cloudflare sandbox execution decoded each aiohttp byte chunk as UTF-8 before passing it to the SSE line decoder. A multi-byte UTF-8 code point split across network chunks therefore raised
UnicodeDecodeErrorinstead of being decoded after the bytes were reassembled.The network stream boundary was incorrectly treated as a text boundary even though UTF-8 character boundaries are independent of aiohttp byte-chunk boundaries. This change lets
_SSELineDecoderaccept byte chunks and decode only complete SSE lines with its incremental decoder, passes raw aiohttp chunks into the decoder, and adds deterministic regression coverage for split UTF-8 code points.No public API signatures change. SSE framing, CR/LF handling, event dispatch, and error propagation remain unchanged for complete text and byte chunks.
Test plan
pytest -q tests/extensions/sandbox/test_cloudflare.py -k 'utf8_split or sse_line_decoder'— 12 passed, 84 deselected.ruff check src/agents/extensions/sandbox/cloudflare/sandbox.py tests/extensions/sandbox/test_cloudflare.py— passed.ruff format --check src/agents/extensions/sandbox/cloudflare/sandbox.py tests/extensions/sandbox/test_cloudflare.py— passed.pyright src/agents/extensions/sandbox/cloudflare/sandbox.py— 0 errors, 0 warnings./usr/bin/env -u OPENAI_API_KEY OPENAI_AGENTS_TEST_IN_CODEX_SANDBOX=1 UV_DEFAULT_INDEX=https://pypi.org/simple bash .agents/skills/code-change-verification/scripts/run.sh— passed (make format,make lint,make typecheck, andmake tests).Issue number
N/A — independently reproduced bug with no pre-existing issue.
Checks
.agents/skills/code-change-verification/scripts/run.sh