VS Code extension: invalid_encrypted_content bricks conversation thread

What version of Codex is running?

0.4.51 (Codex extension for VS Code)

What subscription do you have?

Plus

Which model were you using?

gpt-5.2

What platform is your computer?

Linux 6.6.87.2-microsoft-standard-WSL2 x86_64 x86_64

What issue are you seeing?

Summary

In the Codex VS Code extension, a specific conversation/thread becomes permanently unusable: any further prompt fails with HTTP 400 and
invalid_encrypted_content / "The encrypted content gAAA... could not be verified."
Other conversations still work.

I did NOT switch auth methods (only "Sign in with ChatGPT"), did not touch API keys, and did not change accounts.

Actual behavior

The conversation is "bricked": every attempt to continue fails with:

• invalid_encrypted_content
• "The encrypted content gAAA... could not be verified."

The UI shows repeated "Failed" entries and retry/resume attempts do not recover.
Starting a new conversation works.

Frequency

Happened to me in at least two different conversations.
Seems correlated with long execution/streaming duration (GPT-5.2 Extra high).

Environment

• OS: Windows 11 host + WSL2 (please see uname/os-release below)
• WSL distro: (fill) cat /etc/os-release
• Kernel: (fill) uname -a
• VS Code version: (fill)
• Codex extension version: (fill)
• Auth method: Sign in with ChatGPT (no API key usage)
• Model: GPT-5.2
• Reasoning effort: Extra high

Logs / artifacts

Screenshot of the error in the Codex panel:

<img width="327" height="139" alt="Image" src="https://github.com/user-attachments/assets/594e6150-c634-42bc-9b62-97623253cf42" />

Related issues

• https://github.com/openai/codex/issues/3942
• https://github.com/openai/codex/issues/4409
• https://github.com/openai/codex/issues/3239

What steps can reproduce the bug?

1. Open VS Code with Codex extension (running in WSL2).
2. Sign in using "Sign in with ChatGPT".
3. Start a conversation using model GPT-5.2 and reasoning effort "Extra high".
4. Run a task that takes a long time / involves long streaming (10–30+ minutes), possibly with retries/resume attempts.
5. After the failure occurs once, attempt to continue the same conversation.

What is the expected behavior?

The conversation continues normally, or the extension recovers gracefully (e.g., drops stale state / re-syncs) and keeps the thread usable.

Additional information

_No response_