Troubleshoot repository connections
Find the stage that failed before changing settings. Provider authorization, repository selection, code import and documentation generation have different failure causes.
| Symptom | First checks |
|---|---|
| Repository list is empty | Correct provider account; access to the organization/workspace/project; authorization approval. |
| One private repository is missing | Access granted for that specific repository, including GitHub App installation selection where applicable. |
| Branch cannot be imported | Exact branch name and whether it exists in the selected repository. |
| A previously working import fails | Token expiry/revocation, changed provider permissions, or repository relocation. |
| GitLab self-hosted authorization fails | Instance URL, HTTPS reachability, OAuth callback and scopes. |
| Azure DevOps project is absent | Correct organization/project and token identity permissions. |
A safe recovery sequence
- Read the repository's visible status and error.
- Confirm access in the provider using the same account that authorized DocuWriter.
- Correct the connection or branch, then retry the failed step.
- Wait for a terminal import result before starting another attempt.
A repository that is still synchronizing or indexing is not ready yet. A generation failure after a successful import belongs in generation troubleshooting.
No Autopilot suggestions
A provider connection alone does not enable maintenance for every Space. Check the Space's Autopilot link, branch, triggers and path filters. Activity can show a received event that legitimately produced no suggestions.
Contact support
Include the provider, repository name, branch, approximate failure time and visible error. Include the Space name if there is one. Do not send tokens, OAuth secrets or unredacted authorization headers.
Updated