Troubleshooting
Common issues
Provider request fails
Checklist:
Verify the target provider row exists and is enabled in AI Foundation → AI Providers.
Run Test connection from the AI Providers drawer.
Verify model id and endpoint URL on the provider row (custom/OpenAI-compatible).
Review
sys_logentries written byAiLogService.
HTTP 401 / 403 when fetching protected URL
Checklist:
Enable
basicAuthEnabled.Set both
basicAuthUsernameandbasicAuthPassword.Re-test URL using
HttpAuthUtility.
No statistics shown
Checklist:
Ensure OpenAI key is configured (admin key preferred for org usage endpoint).
Check if API rate limit was hit.
Force refresh in consumer module (if available) or clear cache.
Confirm cache
nst3af_statisticsis available.
Unexpected model behavior
Confirm the default provider in AI Providers matches the feature you expect.
Confirm per-provider model id on the provider row.
Validate temperature and token settings on the provider row.
Operational guidance
Use separate API keys per environment where possible.
Rotate keys after staff/vendor access changes.
Monitor usage trends and adjust limits before cost spikes.
Support information
When escalating an issue, provide:
TYPO3 version
extension version
provider name and model
timestamp and scope/module
sanitized error message from logs