Repeated HTTP 422 cybersecurity-risk flags block legitimate local flaky-test debugging #5335

Describe the bug

GitHub Copilot CLI repeatedly rejects requests to continue investigating an intermittent test crash in my local development workspace.

The task involves reproducing an existing failure, examining debugger output, checking stack and buffer lifetimes, and validating a minimal reliability fix. It does not involve exploit development, unauthorized access, or targeting external systems.

The investigation had already progressed and captured a stack-canary failure. Subsequent continuation requests receive HTTP 422 errors, including after explicitly clarifying the scope and rephrasing the request.

This is a suspected false positive. I cannot determine whether the rejection is triggered by the current prompt, accumulated conversation context, or tool output.

Affected version

• Client: GitHub Copilot CLI 1.0.105
• Model: GPT-6 Astra ( gpt-6-astra )
• OS: Linux x86_64 under WSL2
• Session: Existing, resumed debugging session
• Observed: Repeatedly on November 2, 2026

Steps to reproduce the behavior

1. Continue an existing Copilot CLI session investigating a local C test crash.
2. Ask the assistant to analyze the captured failure and validate an evidence-supported fix.
3. Observe the HTTP 422 rejection.
4. Clarify that the work is authorized local debugging and retry; the rejection persists.

Representative prompt, with internal identifiers omitted:

Please continue diagnosing an intermittent POSIX test failure in my local development workspace. Start from the captured stack-canary failure. Inspect stack and buffer lifetimes to determine whether use-after-free or another memory-management defect caused it. Preserve unrelated changes, make only evidence-supported fixes, and validate against the existing tests.

Expected behavior

Permitted local software-debugging requests should proceed under the applicable safeguards.

Actual Behavior

The CLI repeatedly reports:

Execution failed: 422 This content was flagged for possible cybersecurity risk.
If this seems wrong, try rephrasing your request.
If you’re doing authorized security work that requires more cyber permissive
safeguards, apply for Daybreak access via
https://platform.openai.com/settings/organization/status-and-access before retrying.

Additional context

The repeated rejection interrupts an ongoing reliability investigation and prevents consistent use of the assistant. Please review these requests for false-positive classification and clarify the appropriate support or access-review route for Copilot-managed sessions.