At a glance
| User-visible path | Production behavior | Verified limit |
|---|---|---|
| Git Connector | A GitHub App scans approved documentation paths when an administrator selects Scan by request. It records jobs plus terminology findings. | It defaults off. Each scan allows 50 files, 512 KiB per file, and 2 MiB total. |
| Offline Manifest Sync | An administrator manually imports term-only JSON for report-only comparison. | It defaults off. Each manifest allows 256 KiB and 1,000 terms. |
| Upload controlled evidence | The enterprise workspace creates artifact and revision evidence from an approved file. | A current enterprise freeze, a Retention class and Retention until value, a supported type, and a 25 MiB maximum apply. The artifact record stores both retention values; nothing enforces them. |
| Generic approved HTTPS | Provider-neutral code validates configuration metadata. | Its UI component is not mounted. The manifest permits backend egress only to GitHub. |
| Named candidate profiles | Requirements Yogi, Sodius, and Systems Engineering Studio define candidate data shapes. | They have no enabled provider runtime or user-visible connection path. |
How it works
- Open Settings and find Integration Postures.
- Enable the GitHub connection or manual term-only manifest upload when required.
- For GitHub, open Git Connector and complete Connection configuration.
- Set Allowed paths, select a Confluence space, then use Save connection.
- Use Scan by request. The production GitHub App sends no webhook events, so scans run only on request.
- For offline review, open Offline Manifest Sync and select Import manifest.
- For files, open Terminology Baseline and find Upload controlled evidence.
- Set Retention class and Retention until, then select Upload evidence.
Example
A team enables Git Connector for one approved repository and limits scanning to docs/. Production records jobs and terminology findings. It does not enable Sodius or Requirements Yogi. A separate administrator can upload a controlled PDF into the current enterprise baseline.
What this does not mean
Offline Manifest Sync is a manual import, not a live connector. It does not change glossary terms or create a repository connection.
Upload controlled evidence is an ingestion path, not third-party system access. CAD files produce metadata-only semantic status. PDF parsing records structure metadata.
The visible GitHub workflow is report-only terminology scanning. A separate GitHub enterprise-evidence mapper exists in domain code, but the production scan worker does not call it.
Confluence ingestion is implemented through Forge product APIs. It is a native source path, not an external provider connection.
Related questions
Next step
Evaluate integrations by their current UI, resolver wiring, and manifest permissions.
View on Atlassian Marketplace