CI workflows should run tests only on Linux (ubuntu-latest).
Do not generate any macOS or Windows jobs.

The job failed because the script tried to check if it was running on Linux with:

if [ "Linux" != "Linux" ]; then
  echo "SHOULD_CONTINUE=false" >> $GITHUB_ENV
  echo "Linux is not currently supported."
  exit 1
fi

However, this check will always be false and thus won’t execute its error logic, so that's not the root cause. The critical error occurs later:

Error Failed to get response from the AI model; retried 5 times (total retry wait time: 99.94 seconds)
ERROR: {
  "name": "Error",
  "text": "Failed to get response from the AI model; retried 5 times (total retry wait time: 99.94711389663328 seconds)",
  ...
}

This means the job failed because a request to the AI model (likely GitHub Copilot or a related service) timed out or was blocked by a content filtering policy.

Solution Steps

1. Check for Required Access/Rate Limits:

   • Ensure the GitHub Actions runner has network access to api.githubcopilot.com and any credentials/tokens used are still valid and permitted.
   • You may be hitting rate limits or restrictions from GitHub Copilot/server.

2. Review Content for Filtering:

   • Logs indicate: "Output blocked by content filtering policy", so the payload or request may violate API's acceptable content policies. Audit your prompts or the data sent for accidental inclusion of restricted content.

3. Improve Error Handling/Retry Logic:

   • If the workflow halts immediately on 5 failed attempts, consider exponential backoff and logging sample requests for local debugging.

4. Workflow Recommendations:

   • If you're using a step calling an AI model or Copilot agent:
     • Check the input payloads.
     • Validate against any new anti-abuse/usage policies enforced by GitHub.
     • Contact GitHub support if the request follows policy but is being erroneously blocked.

Code Suggestion Example

If possible, update your workflow step to handle these errors more gracefully and to provide debug information:

- name: Run AI Model Step
  run: |
    for i in {1..10}; do
      response=$(your-ai-call-cmd)
      if [[ $? -eq 0 ]]; then
        echo "AI call succeeded."
        break
      else
        echo "Attempt $i failed. Retrying in $((i*2))s."
        sleep $((i*2))
      fi
      if [[ $i -eq 10 ]]; then
        echo "Final AI call failed, aborting."
        exit 1
      fi
    done

Summary:
Your job fails due to repeated unsuccessful attempts to get a response from the AI model, likely caused by network, credential, or content policy issues. Review credentials, ensure compliance with API usage/content policy, and enhance retry/error handling in your workflow. If the issue persists and you believe there’s no violation, contact GitHub support with log excerpts.

The job failed because repeated POST requests to the GitHub Copilot API returned 400 errors: "Output blocked by content filtering policy." This means the requests your workflow is making to the Copilot or AI API contain content (in the prompt or input payload) that violates GitHub’s content filtering rules, or the API rejected the request body as invalid.

Solution

1. Review AI Prompt/Input for Blocked Content

• Check any prompts, code snippets, or data being sent to the Copilot API in your scripts or workflow. Look for:
  • Sensitive data: credentials, secrets, or personal info.
  • Profane, toxic, or restricted language.
  • Large payloads or malformed JSON.

Code example (find the relevant snippet in your workflow or script that prepares the Copilot/AI API call):

# Example: double-check this content is safe
AI_PAYLOAD=$(cat <<EOF
{
  "question": "$USER_PROMPT"
}
EOF
)

Audit variables (like $USER_PROMPT above) for unexpected or inappropriate content.

2. Improve Error Handling and Retry Logic

Your workflow retries five times, but with each attempt, it's sending the same (likely invalid) payload. Enhance logging to display the actual payload sent (with secrets redacted) and stop retrying if a 400 "blocked by content filtering policy" error occurs—it’s not recoverable by retry.

Modify your script with better error handling:

for i in {1..5}; do
  response=$(curl ... )
  if echo "$response" | grep -q "blocked by content filtering policy"; then
    echo "Prompt blocked. Review content: $AI_PAYLOAD"
    exit 1
  fi
  # Rest of your error/exit logic...
done

3. Validate API Credentials

A 400 error may also indicate malformed authentication headers. Double-check how tokens or headers are passed to the API.

4. Reduce Prompt Complexity/Length

If you’re sending very large prompts, try simplifying or chunking the prompt to reduce issues from exceeding the model’s input constraints.

Summary

• Audit the payload sent to the AI/Copilot API for content or formatting violations.
• Add specific error checks for "blocked by content filtering policy" to fail fast.
• Sanitize and validate prompts and any dynamic string data before API calls.
• Improve error logs to capture and diagnose the actual payload content.

Apply these changes to the script or workflow that initiates the failing API call. If you need help finding the exact file or code, let me know the path or share your AI workflow step, and I can provide more targeted code changes.

Pull request overview

This PR establishes a comprehensive OSS-ready repository structure for qumo, a Media over QUIC relay/CDN project. It includes all necessary templates, workflows, and organizational files to support open source development while maintaining the project as a stub/skeleton without actual implementation.

Key changes:

• Created standard Go project structure with /cmd, /internal, /pkg, /docs, /configs, and /monitoring directories
• Added comprehensive GitHub automation including CI/CD workflows, issue templates, and dependency management
• Established community health files (Code of Conduct, Contributing guidelines, PR template)

Reviewed changes

Copilot reviewed 20 out of 21 changed files in this pull request and generated 7 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.